use optimization level 3 and thin LTO for release builds (#1405)

Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/1405
Co-authored-by: Alex Auvolat <lx@deuxfleurs.fr>
Co-committed-by: Alex Auvolat <lx@deuxfleurs.fr>

.cargo/config.toml

@@ -1,3 +0,0 @@
-[target.x86_64-unknown-linux-gnu]
-linker = "clang"
-rustflags = ["-C", "link-arg=-fuse-ld=mold"]

.woodpecker/debug.yaml

@@ -1,3 +1,6 @@
+labels:
+  nix: "enabled"
+
 when:
   event:
   - push
@@ -9,39 +12,33 @@ when:
 
 steps:
   - name: check formatting
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
     commands:
-      - nix-shell --attr devShell --run "cargo fmt -- --check"
+      - nix-build -j4 --attr flakePackages.fmt
 
   - name: build
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
     commands:
-      - nix-build --no-build-output --attr clippy.amd64 --argstr git_version ${CI_COMMIT_TAG:-$CI_COMMIT_SHA}
+      - nix-build -j4 --attr flakePackages.dev
 
-  - name: unit + func tests
+  - name: unit + func tests (lmdb)
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
-    environment:
-      GARAGE_TEST_INTEGRATION_EXE: result-bin/bin/garage
-      GARAGE_TEST_INTEGRATION_PATH: tmp-garage-integration
     commands:
-      - nix-build --no-build-output --attr clippy.amd64 --argstr git_version ${CI_COMMIT_TAG:-$CI_COMMIT_SHA}
+      - nix-build -j4 --attr flakePackages.tests-lmdb
-      - nix-build --no-build-output --attr test.amd64
+
-      - ./result/bin/garage_db-*
+  - name: unit + func tests (sqlite)
-      - ./result/bin/garage_api-*
+    image: nixpkgs/nix:nixos-24.05
-      - ./result/bin/garage_model-*
+    commands:
-      - ./result/bin/garage_rpc-*
+      - nix-build -j4 --attr flakePackages.tests-sqlite
-      - ./result/bin/garage_table-*
+
-      - ./result/bin/garage_util-*
+  - name: unit + func tests (fjall)
-      - ./result/bin/garage_web-*
+    image: nixpkgs/nix:nixos-24.05
-      - ./result/bin/garage-*
+    commands:
-      - GARAGE_TEST_INTEGRATION_DB_ENGINE=lmdb ./result/bin/integration-* || (cat tmp-garage-integration/stderr.log; false)
+      - nix-build -j4 --attr flakePackages.tests-fjall
-      - nix-shell --attr ci --run "killall -9 garage" || true
-      - GARAGE_TEST_INTEGRATION_DB_ENGINE=sqlite ./result/bin/integration-* || (cat tmp-garage-integration/stderr.log; false)
-      - rm result
-      - rm -rv tmp-garage-integration
 
   - name: integration tests
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
     commands:
-      - nix-build --no-build-output --attr clippy.amd64 --argstr git_version ${CI_COMMIT_TAG:-$CI_COMMIT_SHA}
+      - nix-build -j4 --attr flakePackages.dev
       - nix-shell --attr ci --run ./script/test-smoke.sh || (cat /tmp/garage.log; false)
+    depends_on: [ build ]

.woodpecker/publish.yaml

@@ -1,3 +1,6 @@
+labels:
+  nix: "enabled"
+
 when:
   event:
   - deployment
@@ -8,20 +11,21 @@ depends_on:
 
 steps:
   - name: refresh-index
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
-    secrets:
+    environment:
-      - source: garagehq_aws_access_key_id
+      AWS_ACCESS_KEY_ID:
-        target: AWS_ACCESS_KEY_ID
+        from_secret: garagehq_aws_access_key_id
-      - source: garagehq_aws_secret_access_key
+      AWS_SECRET_ACCESS_KEY:
-        target: AWS_SECRET_ACCESS_KEY
+        from_secret: garagehq_aws_secret_access_key
     commands:
       - mkdir -p /etc/nix && cp nix/nix.conf /etc/nix/nix.conf
       - nix-shell --attr ci --run "refresh_index"
 
   - name: multiarch-docker
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
-    secrets:
+    environment:
-      - docker_auth
+      DOCKER_AUTH:
+        from_secret: docker_auth
     commands:
       - mkdir -p /root/.docker
       - echo $DOCKER_AUTH > /root/.docker/config.json

.woodpecker/release.yaml

@@ -1,3 +1,6 @@
+labels:
+  nix: "enabled"
+
 when:
   event:
   - deployment
@@ -16,18 +19,17 @@ matrix:
 
 steps:
   - name: build
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
     commands:
-      - nix-build --no-build-output --attr pkgs.${ARCH}.release --argstr git_version ${CI_COMMIT_TAG:-$CI_COMMIT_SHA}
+      - nix-build --attr releasePackages.${ARCH} --argstr git_version ${CI_COMMIT_TAG:-$CI_COMMIT_SHA}
 
   - name: check is static binary
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
     commands:
-      - nix-build --no-build-output --attr pkgs.${ARCH}.release --argstr git_version ${CI_COMMIT_TAG:-$CI_COMMIT_SHA}
+      - nix-shell --attr ci --run "./script/not-dynamic.sh result/bin/garage"
-      - nix-shell --attr ci --run "./script/not-dynamic.sh result-bin/bin/garage"
 
   - name: integration tests
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
     commands:
       - nix-shell --attr ci --run ./script/test-smoke.sh || (cat /tmp/garage.log; false)
     when:
@@ -37,7 +39,7 @@ steps:
           ARCH: i386
 
   - name: upgrade tests
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
     commands:
       - nix-shell --attr ci --run "./script/test-upgrade.sh v0.8.4 x86_64-unknown-linux-musl" || (cat /tmp/garage.log; false)
     when:
@@ -45,24 +47,23 @@ steps:
           ARCH: amd64
 
   - name: push static binary
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
     environment:
       TARGET: "${TARGET}"
-    secrets:
+      AWS_ACCESS_KEY_ID:
-      - source: garagehq_aws_access_key_id
+        from_secret: garagehq_aws_access_key_id
-        target: AWS_ACCESS_KEY_ID
+      AWS_SECRET_ACCESS_KEY:
-      - source: garagehq_aws_secret_access_key
+        from_secret: garagehq_aws_secret_access_key
-        target: AWS_SECRET_ACCESS_KEY
     commands:
       - nix-shell --attr ci --run "to_s3"
 
   - name: docker build and publish
-    image: nixpkgs/nix:nixos-22.05
+    image: nixpkgs/nix:nixos-24.05
     environment:
       DOCKER_PLATFORM: "linux/${ARCH}"
       CONTAINER_NAME: "dxflrs/${ARCH}_garage"
-    secrets:
+      DOCKER_AUTH:
-      - docker_auth
+        from_secret: docker_auth
     commands:
       - mkdir -p /root/.docker
       - echo $DOCKER_AUTH > /root/.docker/config.json

Cargo.toml

@@ -8,7 +8,10 @@ members = [
 	"src/table",
 	"src/block",
 	"src/model",
-	"src/api",
+	"src/api/common",
+	"src/api/s3",
+	"src/api/k2v",
+	"src/api/admin",
 	"src/web",
 	"src/garage",
 	"src/k2v-client",
@@ -21,15 +24,18 @@ default-members = ["src/garage"]
 
 # Internal Garage crates
 format_table = { version = "0.1.1", path = "src/format-table" }
-garage_api = { version = "1.0.0", path = "src/api" }
+garage_api_common = { version = "1.3.1", path = "src/api/common" }
-garage_block = { version = "1.0.0", path = "src/block" }
+garage_api_admin = { version = "1.3.1", path = "src/api/admin" }
-garage_db = { version = "1.0.0", path = "src/db", default-features = false }
+garage_api_s3 = { version = "1.3.1", path = "src/api/s3" }
-garage_model = { version = "1.0.0", path = "src/model", default-features = false }
+garage_api_k2v = { version = "1.3.1", path = "src/api/k2v" }
-garage_net = { version = "1.0.0", path = "src/net" }
+garage_block = { version = "1.3.1", path = "src/block" }
-garage_rpc = { version = "1.0.0", path = "src/rpc" }
+garage_db = { version = "1.3.1", path = "src/db", default-features = false }
-garage_table = { version = "1.0.0", path = "src/table" }
+garage_model = { version = "1.3.1", path = "src/model", default-features = false }
-garage_util = { version = "1.0.0", path = "src/util" }
+garage_net = { version = "1.3.1", path = "src/net" }
-garage_web = { version = "1.0.0", path = "src/web" }
+garage_rpc = { version = "1.3.1", path = "src/rpc" }
+garage_table = { version = "1.3.1", path = "src/table" }
+garage_util = { version = "1.3.1", path = "src/util" }
+garage_web = { version = "1.3.1", path = "src/web" }
 k2v-client = { version = "0.0.4", path = "src/k2v-client" }
 
 # External crates from crates.io
@@ -46,21 +52,19 @@ chrono = "0.4"
 crc32fast = "1.4"
 crc32c = "0.6"
 crypto-common = "0.1"
-digest = "0.10"
-err-derive = "0.3"
 gethostname = "0.4"
 git-version = "0.3.4"
 hex = "0.4"
 hexdump = "0.1"
 hmac = "0.12"
-idna = "0.5"
 itertools = "0.12"
 ipnet = "2.9.0"
 lazy_static = "1.4"
 md-5 = "0.10"
 mktemp = "0.5"
-nix = { version = "0.27", default-features = false, features = ["fs"] }
+nix = { version = "0.29", default-features = false, features = ["fs"] }
 nom = "7.1"
+parking_lot = "0.12"
 parse_duration = "2.1"
 pin-project = "1.0.12"
 pnet_datalink = "0.34"
@@ -79,12 +83,14 @@ pretty_env_logger = "0.5"
 structopt = { version = "0.3", default-features = false }
 syslog-tracing = "0.3"
 tracing = "0.1"
+tracing-journald = "0.3.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
 
 heed = { version = "0.11", default-features = false, features = ["lmdb"] }
-rusqlite = "0.31.0"
+rusqlite = "0.37"
 r2d2 = "0.8"
-r2d2_sqlite = "0.24"
+r2d2_sqlite = "0.31"
+fjall = "2.4"
 
 async-compression = { version = "0.4", features = ["tokio", "zstd"] }
 zstd = { version = "0.13", default-features = false }
@@ -127,26 +133,21 @@ opentelemetry-contrib = "0.9"
 prometheus = "0.13"
 
 # used by the k2v-client crate only
-aws-sigv4 = { version = "1.1" }
+aws-sigv4 = { version = "1.1", default-features = false }
-hyper-rustls = { version = "0.26", features = ["http2"] }
+hyper-rustls = { version = "0.26", default-features = false, features = ["http1", "http2", "ring", "rustls-native-certs"] }
 log = "0.4"
-thiserror = "1.0"
+thiserror = "2.0"
 
 # ---- used only as build / dev dependencies ----
 assert-json-diff = "2.0"
 rustc_version = "0.4.0"
 static_init = "1.0"
-
+aws-smithy-runtime = { version = "1.8", default-features = false, features = ["tls-rustls"] }
-aws-config = "1.1.4"
+aws-sdk-config = { version = "1.62", default-features = false }
-aws-sdk-config = "1.13"
+aws-sdk-s3 = { version = "1.79", default-features = false, features = ["rt-tokio"] }
-aws-sdk-s3 = "1.14"
-
-[profile.dev]
-#lto = "thin"	# disabled for now, adds 2-4 min to each CI build
-lto = "off"
 
 [profile.release]
-lto = true
+lto = "thin"
-codegen-units = 1
+codegen-units = 16
-opt-level = "s"
+opt-level = 3
-strip = true
+strip = "debuginfo"

Dockerfile

@@ -3,5 +3,5 @@ FROM scratch
 ENV RUST_BACKTRACE=1
 ENV RUST_LOG=garage=info
 
-COPY result-bin/bin/garage /
+COPY result/bin/garage /
 CMD [ "/garage", "server"]

Makefile

@@ -1,13 +1,8 @@
-.PHONY: doc all release shell run1 run2 run3
+.PHONY: doc all run1 run2 run3
 
 all:
-	clear; cargo build
+	clear
-
+	cargo build
-release:
-	nix-build --attr pkgs.amd64.release --no-build-output
-
-shell:
-	nix-shell
 
 # ----
 

default.nix

@@ -3,53 +3,22 @@
 with import ./nix/common.nix;
 
 let
-  pkgs = import pkgsSrc { };
+  pkgs = import nixpkgs { };
   compile = import ./nix/compile.nix;
 
-  build_debug_and_release = (target: {
+  build_release = target: (compile {
-    debug = (compile {
+    inherit target system git_version nixpkgs;
-      inherit system target git_version pkgsSrc cargo2nixOverlay;
+    crane = flake.inputs.crane;
-      release = false;
+    rust-overlay = flake.inputs.rust-overlay;
-    }).workspace.garage { compileMode = "build"; };
+    release = true;
-
+  }).garage;
-    release = (compile {
-      inherit system target git_version pkgsSrc cargo2nixOverlay;
-      release = true;
-    }).workspace.garage { compileMode = "build"; };
-  });
-
-  test = (rustPkgs:
-    pkgs.symlinkJoin {
-      name = "garage-tests";
-      paths =
-        builtins.map (key: rustPkgs.workspace.${key} { compileMode = "test"; })
-        (builtins.attrNames rustPkgs.workspace);
-    });
 
 in {
-  pkgs = {
+  releasePackages = {
-    amd64 = build_debug_and_release "x86_64-unknown-linux-musl";
+    amd64 = build_release "x86_64-unknown-linux-musl";
-    i386 = build_debug_and_release "i686-unknown-linux-musl";
+    i386 = build_release "i686-unknown-linux-musl";
-    arm64 = build_debug_and_release "aarch64-unknown-linux-musl";
+    arm64 = build_release "aarch64-unknown-linux-musl";
-    arm = build_debug_and_release "armv6l-unknown-linux-musleabihf";
+    arm = build_release "armv6l-unknown-linux-musleabihf";
-  };
-  test = {
-    amd64 = test (compile {
-      inherit system git_version pkgsSrc cargo2nixOverlay;
-      target = "x86_64-unknown-linux-musl";
-      features = [
-        "garage/bundled-libs"
-        "garage/k2v"
-        "garage/lmdb"
-        "garage/sqlite"
-      ];
-    });
-  };
-  clippy = {
-    amd64 = (compile {
-      inherit system git_version pkgsSrc cargo2nixOverlay;
-      target = "x86_64-unknown-linux-musl";
-      compiler = "clippy";
-    }).workspace.garage { compileMode = "build"; };
   };
+  flakePackages = flake.packages.${system};
 }

doc/api/garage-admin-v1.yml

@@ -687,7 +687,7 @@ paths:
       operationId: "GetBucketInfo"
       summary: "Get a bucket"
       description: |
-        Given a bucket identifier (`id`) or a global alias (`alias`), get its information.
+        Given a bucket identifier (`id`) or a global alias (`globalAlias`), get its information.
         It includes its aliases, its web configuration, keys that have some permissions
         on it, some statistics (number of objects, size), number of dangling multipart uploads,
         and its quotas (if any).
@@ -701,7 +701,7 @@ paths:
           example: "b4018dc61b27ccb5c64ec1b24f53454bbbd180697c758c4d47a22a8921864a87"
           schema:
             type: string
-        - name: alias
+        - name: globalAlias
           in: query
           description: |
             The exact global alias of one of the existing buckets.

doc/book/build/python.md

@@ -23,7 +23,7 @@ client = minio.Minio(
   "GKyourapikey",
   "abcd[...]1234",
   # Force the region, this is specific to garage
-  region="region",
+  region="garage",
 )
 ```
 

doc/book/connect/apps/index.md

@@ -12,7 +12,7 @@ In this section, we cover the following web applications:
 | [Mastodon](#mastodon)     | ✅       | Natively supported    |
 | [Matrix](#matrix)     | ✅       |  Tested with `synapse-s3-storage-provider`    |
 | [ejabberd](#ejabberd)     | ✅       |  `mod_s3_upload`    |
-| [Pixelfed](#pixelfed)     | ❓       |  Not yet tested    |
+| [Pixelfed](#pixelfed)     | ✅       |  Natively supported    |
 | [Pleroma](#pleroma)     | ❓       |  Not yet tested    |
 | [Lemmy](#lemmy)     | ✅        |  Supported with pict-rs    |
 | [Funkwhale](#funkwhale)     | ❓       | Not yet tested     |
@@ -69,7 +69,7 @@ $CONFIG = array(
         'hostname' => '127.0.0.1', // Can also be a domain name, eg. garage.example.com
         'port' => 3900,            // Put your reverse proxy port or your S3 API port
         'use_ssl' => false,        // Set it to true if you have a TLS enabled reverse proxy
-        'region' => 'garage',      // Garage has only one region named "garage"
+        'region' => 'garage',      // Garage default region is named "garage", edit according to your cluster config
         'use_path_style' => true   // Garage supports only path style, must be set to true
     ],
 ],
@@ -135,7 +135,7 @@ bucket but doesn't also know the secret encryption key.
 *Click on the picture to zoom*
 
 Add a new external storage. Put what you want in "folder name" (eg. "shared"). Select "Amazon S3". Keep "Access Key" for the Authentication field.
-In Configuration, put your bucket name (eg. nextcloud), the host (eg. 127.0.0.1), the port (eg. 3900 or 443), the region (garage). Tick the SSL box if you have put an HTTPS proxy in front of garage. You must tick the "Path access" box and you must leave the "Legacy authentication (v2)" box empty. Put your Key ID (eg. GK...) and your Secret Key in the last two input boxes. Finally click on the tick symbol on the right of your screen.
+In Configuration, put your bucket name (eg. nextcloud), the host (eg. 127.0.0.1), the port (eg. 3900 or 443), the region ("garage" if you use the default, or the one your configured in your `garage.toml`). Tick the SSL box if you have put an HTTPS proxy in front of garage. You must tick the "Path access" box and you must leave the "Legacy authentication (v2)" box empty. Put your Key ID (eg. GK...) and your Secret Key in the last two input boxes. Finally click on the tick symbol on the right of your screen.
 
 Now go to your "Files" app and a new "linked folder" has appeared with the name you chose earlier (eg. "shared").
 
@@ -191,10 +191,10 @@ garage key create peertube-key
 
 Keep the Key ID and the Secret key in a pad, they will be needed later.  
 
-We need two buckets, one for normal videos (named peertube-video) and one for webtorrent videos (named peertube-playlist).
+We need two buckets, one for normal videos (named peertube-videos) and one for webtorrent videos (named peertube-playlists).
 ```bash
 garage bucket create peertube-videos
-garage bucket create peertube-playlist
+garage bucket create peertube-playlists
 ```
 
 Now we allow our key to read and write on these buckets:
@@ -238,7 +238,7 @@ object_storage:
   # Put localhost only if you have a garage instance running on that node
   endpoint: 'http://localhost:3900' # or "garage.example.com" if you have TLS on port 443
 
-  # Garage supports only one region for now, named garage
+  # Garage default region is named "garage", edit according to your config
   region: 'garage'
 
   credentials:
@@ -253,7 +253,7 @@ object_storage:
     proxify_private_files: false
 
   streaming_playlists:
-    bucket_name: 'peertube-playlist'
+    bucket_name: 'peertube-playlists'
 
     # Keep it empty for our example
     prefix: ''
@@ -335,6 +335,7 @@ From the [official Mastodon documentation](https://docs.joinmastodon.org/admin/t
 
 ```bash
 $ RAILS_ENV=production bin/tootctl media remove --days 3
+$ RAILS_ENV=production bin/tootctl media remove --days 15 --prune-profiles
 $ RAILS_ENV=production bin/tootctl media remove-orphans
 $ RAILS_ENV=production bin/tootctl preview_cards remove --days 15
 ```
@@ -353,8 +354,6 @@ Imports:	1.7 KB
 Settings:	0 Bytes
 ```
 
-Unfortunately, [old avatars and headers cannot currently be cleaned up](https://github.com/mastodon/mastodon/issues/9567).
-
 ### Migrating your data
 
 Data migration should be done with an efficient S3 client.
@@ -442,7 +441,7 @@ media_storage_providers:
   store_synchronous: True  # do we want to wait that the file has been written before returning?
   config:
     bucket: matrix       # the name of our bucket, we chose matrix earlier
-    region_name: garage  # only "garage" is supported for the region field
+    region_name: garage  # "garage" by default, edit according to your cluster config
     endpoint_url: http://localhost:3900 # the path to the S3 endpoint
     access_key_id: "GKxxx" # your Key ID
     secret_access_key: "xxxx" # your Secret Key

doc/book/connect/backup.md

@@ -161,3 +161,49 @@ kopia repository validate-provider
 
 You can then run all the standard kopia commands: `kopia snapshot create`, `kopia mount`...
 Everything should work out-of-the-box.
+
+## Plakar
+
+Create your key and bucket on Garage server:
+
+```bash
+garage key create my-plakar-key
+garage bucket create plakar-backups
+garage bucket allow plakar-backups --read --write --key my-plakar-key
+…
+```
+
+On Plakar server, add your Garage as a storage location:
+```bash
+plakar store add garageS3 s3://my-garage.tld/plakar-backups \
+region=garage # Or as you've specified in garage.toml \
+access_key=<Key ID from "garage key info my-plakar-key"> \
+secret_access_key=<Secret key from "garage key info my-plakar-key">
+```
+
+Then create the repository.
+```bash
+plakar at @garageS3 create -plaintext # Unencrypted
+# or
+plakar at @garageS3 create #encrypted
+```
+
+If you encrypt your backups (Plakar default), you will need to define a strong passphrase. Do not forget to save your password safely. It will be needed to decrypt your backups.
+
+
+After the repository has been created, check that everything works as expected (that might give an empty result as no file has been added yet, but no error message):
+```bash
+plakar at @garageS3 check
+```
+
+Now that everything is configure, you can use Garage as your backups storage. For instance sync it with a local backup storage:
+```bash
+$ plakar at ~/backups sync to @garageS3
+```
+
+Or list the S3 storage content:
+```bash
+$ plakar at @garageS3 ls
+```
+
+More information in Plakar documentation: https://www.plakar.io/docs/main/quickstart/

doc/book/connect/repositories.md

@@ -17,7 +17,7 @@ Garage can also help you serve this content.
 
 ## Gitea
 
-You can use Garage with Gitea to store your [git LFS](https://git-lfs.github.com/) data, your users' avatar, and their attachements.
+You can use Garage with Gitea to store your [git LFS](https://git-lfs.github.com/) data, your users' avatar, and their attachments.
 You can configure a different target for each data type (check `[lfs]` and `[attachment]` sections of the Gitea documentation) and you can provide a default one through the `[storage]` section.
 
 Let's start by creating a key and a bucket (your key id and secret will be needed later, keep them somewhere):

doc/book/cookbook/ansible.md

@@ -8,18 +8,18 @@ have published Ansible roles.  We list them and compare them below.
 
 ## Comparison of Ansible roles
 
-| Feature                            | [ansible-role-garage](#zorun-ansible-role-garage) | [garage-docker-ansible-deploy](#moan0s-garage-docker-ansible-deploy) |
+| Feature                            | [ansible-role-garage](#zorun-ansible-role-garage) | [garage-docker-ansible-deploy](#moan0s-garage-docker-ansible-deploy) | [eddster ansible-role-garage](#eddster-ansible-role-garage) |
-|------------------------------------|---------------------------------------------|---------------------------------------------------------------|
+|------------------------------------|---------------------------------------------|---------------------------------------------------------------|---------------------------------|
-| **Runtime**                        | Systemd                                     | Docker                                                        |
+| **Runtime**                        | Systemd                                     | Docker                                                        | Systemd                         |
-| **Target OS**                      | Any Linux                                   | Any Linux                                                     |
+| **Target OS**                      | Any Linux                                   | Any Linux                                                     | Any Linux                       |
-| **Architecture**                   | amd64, arm64, i686                          | amd64, arm64                                                  |
+| **Architecture**                   | amd64, arm64, i686                          | amd64, arm64                                                  | arm64, arm, 386, amd64                            |
-| **Additional software**            | None                                        | Traefik                                                       |
+| **Additional software**            | None                                        | Traefik                                                       | Ngnix and Keepalived (optional) |
-| **Automatic node connection**      | ❌                                          | ✅                                                            |
+| **Automatic node connection**      | ❌                                          | ✅                                                            | ✅                              |
-| **Layout management**              | ❌                                          | ✅                                                            |
+| **Layout management**              | ❌                                          | ✅                                                            | ✅                              |
-| **Manage buckets & keys**          | ❌                                          | ✅ (basic)                                                    |
+| **Manage buckets & keys**          | ❌                                          | ✅ (basic)                                                    | ✅                              |
-| **Allow custom Garage config**     | ✅                                          | ❌                                                            |
+| **Allow custom Garage config**     | ✅                                          | ❌                                                            | ❌                              |
-| **Facilitate Garage upgrades**     | ✅                                          | ❌                                                            |
+| **Facilitate Garage upgrades**     | ✅                                          | ❌                                                            | ✅                              |
-| **Multiple instances on one host** | ✅                                          | ✅                                                            |
+| **Multiple instances on one host** | ✅                                          | ✅                                                            | ❌                              |
 
 
 ## zorun/ansible-role-garage
@@ -49,3 +49,15 @@ structured DNS names, etc).
 
 As a result, this role makes it easier to start with Garage on Ansible,
 but is less flexible.
+
+## eddster2309/ansible-role-garage
+
+[Source code](https://github.com/eddster2309/ansible-role-garage), [Ansible galaxy](https://galaxy.ansible.com/ui/standalone/roles/eddster2309/garage/)
+
+This role is a opinionated but customisable role using the official Garage
+static binaries and only requires Systemd. As such it should work on any
+Linux based host. It includes all the nesscary configuration to
+automatically setup a clustered Garage deployment. Most Garage
+configuration options are exposed through Ansible variables so while you
+can't provide a custom config you can get very close. It can optionally
+installed a HA nginx deployment with Keepalived.

doc/book/cookbook/binary-packages.md

@@ -15,9 +15,10 @@ Alpine Linux repositories (available since v3.17):
 apk add garage
 ```
 
-The default configuration file is installed to `/etc/garage.toml`. You can run
+The default configuration file is installed to `/etc/garage/garage.toml`. You can run
-Garage using: `rc-service garage start`. If you don't specify `rpc_secret`, it
+Garage using: `rc-service garage start`.
-will be automatically replaced with a random string on the first start.
+
+If you don't specify `rpc_secret`, it will be automatically replaced with a random string on the first start.
 
 Please note that this package is built without Consul discovery, Kubernetes
 discovery, OpenTelemetry exporter, and K2V features (K2V will be enabled once
@@ -26,7 +27,7 @@ it's stable).
 
 ## Arch Linux
 
-Garage is available in the [AUR](https://aur.archlinux.org/packages/garage).
+Garage is available in the official repositories under [extra](https://archlinux.org/packages/extra/x86_64/garage).
 
 ## FreeBSD
 

doc/book/cookbook/kubernetes.md

@@ -11,7 +11,7 @@ Firstly clone the repository:
 
 ```bash
 git clone https://git.deuxfleurs.fr/Deuxfleurs/garage
-cd garage/scripts/helm
+cd garage/script/helm
 ```
 
 Deploy with default options:
@@ -26,6 +26,13 @@ Or deploy with custom values:
 helm install --create-namespace --namespace garage garage ./garage -f values.override.yaml
 ```
 
+If you want to manage the CustomRessourceDefinition used by garage for its `kubernetes_discovery` outside of the helm chart, add `garage.kubernetesSkipCrd: true` to your custom values and use the kustomization before deploying the helm chart:
+
+```bash
+kubectl apply -k ../k8s/crd
+helm install --create-namespace --namespace garage garage ./garage -f values.override.yaml
+```
+
 After deploying, cluster layout must be configured manually as described in [Creating a cluster layout](@/documentation/quick-start/_index.md#creating-a-cluster-layout). Use the following command to access garage CLI:
 
 ```bash
@@ -86,3 +93,62 @@ helm delete --namespace garage garage
 ```
 
 Note that this will leave behind custom CRD `garagenodes.deuxfleurs.fr`, which must be removed manually if desired.
+
+## Increase PVC size on running Garage instances
+
+Since the Garage Helm chart creates the data and meta PVC based on `StatefulSet` templates, increasing the PVC size can be a bit tricky.
+
+### Confirm the `StorageClass` used for Garage supports volume expansion
+
+Confirm the storage class used for garage.
+
+```bash
+kubectl -n garage get pvc 
+NAME            STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS     VOLUMEATTRIBUTESCLASS   AGE
+data-garage-0   Bound    pvc-080360c9-8ce3-4acf-8579-1701e57b7f3f   30Gi       RWO            longhorn-local   <unset>                 77d
+data-garage-1   Bound    pvc-ab8ba697-6030-4fc7-ab3c-0d6df9e3dbc0   30Gi       RWO            longhorn-local   <unset>                 5d8h
+data-garage-2   Bound    pvc-3ab37551-0231-4604-986d-136d0fd950ec   30Gi       RWO            longhorn-local   <unset>                 5d5h
+meta-garage-0   Bound    pvc-3b457302-3023-4169-846e-c928c5f2ea65   3Gi        RWO            longhorn-local   <unset>                 77d
+meta-garage-1   Bound    pvc-49ace2b9-5c85-42df-9247-51c4cf64b460   3Gi        RWO            longhorn-local   <unset>                 5d8h
+meta-garage-2   Bound    pvc-99e2e50f-42b4-4128-ae2f-b52629259723   3Gi        RWO            longhorn-local   <unset>                 5d5h
+```
+
+In this case, the storage class is `longhorn-local`. Now, check if `ALLOWVOLUMEEXPANSION` is true for the used `StorageClass`.
+
+```bash
+kubectl get storageclasses.storage.k8s.io longhorn-local
+NAME             PROVISIONER          RECLAIMPOLICY   VOLUMEBINDINGMODE   ALLOWVOLUMEEXPANSION   AGE
+longhorn-local   driver.longhorn.io   Delete          Immediate           true                   103d
+```
+
+If your `StorageClass` does not support volume expansion, double check if you can enable it. Otherwise, your only real option is to spin up a new Garage cluster with increased size and migrate all data over.
+
+If your `StorageClass` supports expansion, you are free to continue.
+
+### Increase the size of the PVCs
+
+Increase the size of all PVCs to your desired size.
+
+```bash
+kubectl -n garage edit pvc data-garage-0
+kubectl -n garage edit pvc data-garage-1
+kubectl -n garage edit pvc data-garage-2
+kubectl -n garage edit pvc meta-garage-0
+kubectl -n garage edit pvc meta-garage-1
+kubectl -n garage edit pvc meta-garage-2
+```
+
+### Increase the size of the `StatefulSet` PVC template
+
+This is an optional step, but if not done, future instances of Garage will be created with the original size from the template.
+
+```bash
+kubectl -n garage delete sts --cascade=orphan garage 
+statefulset.apps "garage" deleted
+```
+
+This will remove the Garage `StatefulSet` but leave the pods running. It may seem destructive but needs to be done this way since edits to the size of PVC templates are prohibited.
+
+### Redeploy the `StatefulSet`
+
+Now the size of future PVCs can be increased, and the Garage Helm chart can be upgraded. The new `StatefulSet` should take ownership of the orphaned pods again.

doc/book/cookbook/real-world.md

@@ -96,14 +96,14 @@ to store 2 TB of data in total.
 ## Get a Docker image
 
 Our docker image is currently named `dxflrs/garage` and is stored on the [Docker Hub](https://hub.docker.com/r/dxflrs/garage/tags?page=1&ordering=last_updated).
-We encourage you to use a fixed tag (eg. `v1.0.0`) and not the `latest` tag.
+We encourage you to use a fixed tag (eg. `v1.3.0`) and not the `latest` tag.
-For this example, we will use the latest published version at the time of the writing which is `v1.0.0` but it's up to you
+For this example, we will use the latest published version at the time of the writing which is `v1.3.0` but it's up to you
 to check [the most recent versions on the Docker Hub](https://hub.docker.com/r/dxflrs/garage/tags?page=1&ordering=last_updated).
 
 For example:
 
 ```
-sudo docker pull dxflrs/garage:v1.0.0
+sudo docker pull dxflrs/garage:v1.3.0
 ```
 
 ## Deploying and configuring Garage
@@ -171,7 +171,7 @@ docker run \
   -v /etc/garage.toml:/etc/garage.toml \
   -v /var/lib/garage/meta:/var/lib/garage/meta \
   -v /var/lib/garage/data:/var/lib/garage/data \
-  dxflrs/garage:v1.0.0
+  dxflrs/garage:v1.3.0
 ```
 
 With this command line, Garage should be started automatically at each boot.
@@ -185,7 +185,7 @@ If you want to use `docker-compose`, you may use the following `docker-compose.y
 version: "3"
 services:
   garage:
-    image: dxflrs/garage:v1.0.0
+    image: dxflrs/garage:v1.3.0
     network_mode: "host"
     restart: unless-stopped
     volumes:

doc/book/cookbook/systemd.md

@@ -28,6 +28,7 @@ StateDirectory=garage
 DynamicUser=true
 ProtectHome=true
 NoNewPrivileges=true
+LimitNOFILE=42000
 
 [Install]
 WantedBy=multi-user.target

doc/book/design/goals.md

@@ -50,3 +50,20 @@ locations. They use Garage themselves for the following tasks:
 
 The Deuxfleurs Garage cluster is a multi-site cluster currently composed of
 9 nodes in 3 physical locations.
+
+### Triplebit
+
+[Triplebit](https://www.triplebit.org) is a non-profit hosting provider and
+ISP focused on improving access to privacy-related services. They use
+Garage themselves for the following tasks:
+
+- Hosting of their homepage, [privacyguides.org](https://www.privacyguides.org/), and various other static sites
+
+- As a Mastodon object storage backend for [mstdn.party](https://mstdn.party/) and [mstdn.plus](https://mstdn.plus/)
+
+- As a PeerTube storage backend for [neat.tube](https://neat.tube/)
+
+- As a [Matrix media backend](https://github.com/matrix-org/synapse-s3-storage-provider)
+
+Triplebit's Garage cluster is a multi-site cluster currently composed of
+10 nodes in 3 physical locations.

doc/book/development/devenv.md

@@ -36,7 +36,7 @@ sudo killall nix-daemon
 Now you can enter our nix-shell, all the required packages will be downloaded but they will not pollute your environment outside of the shell:
 
 ```bash
-nix-shell
+nix-shell -A devShell
 ```
 
 You can use the traditional Rust development workflow:
@@ -65,8 +65,8 @@ nix-build -j $(nproc) --max-jobs auto
 ```
 
 Our build has multiple parameters you might want to set:
-  - `release` build with release optimisations instead of debug
+  - `release` to build with release optimisations instead of debug
-  - `target allows` for cross compilation
+  - `target` allows for cross compilation
   - `compileMode` can be set to test or bench to build a unit test runner
   - `git_version` to inject the hash to display when running `garage stats`
 

doc/book/operations/multi-hdd.md

@@ -21,14 +21,14 @@ data_dir = [
 ```
 
 Garage will automatically balance all blocks stored by the node
-among the different specified directories, proportionnally to the
+among the different specified directories, proportionally to the
 specified capacities.
 
 ## Updating the list of storage locations
 
 If you add new storage locations to your `data_dir`,
 Garage will not rebalance existing data between storage locations.
-Newly written blocks will be balanced proportionnally to the specified capacities,
+Newly written blocks will be balanced proportionally to the specified capacities,
 and existing data may be moved between drives to improve balancing,
 but only opportunistically when a data block is re-written (e.g. an object
 is re-uploaded, or an object with a duplicate block is uploaded).

doc/book/operations/recovering.md

@@ -5,7 +5,7 @@ weight = 40
 
 Garage is meant to work on old, second-hand hardware.
 In particular, this makes it likely that some of your drives will fail, and some manual intervention will be needed.
-Fear not! For Garage is fully equipped to handle drive failures, in most common cases.
+Fear not! Garage is fully equipped to handle drive failures, in most common cases.
 
 ## A note on availability of Garage
 
@@ -61,7 +61,7 @@ garage repair -a --yes blocks
 
 This will re-synchronize blocks of data that are missing to the new HDD, reading them from copies located on other nodes.
 
-You can check on the advancement of this process by doing the following command: 
+You can check on the advancement of this process by doing the following command:
 
 ```bash
 garage stats -a

doc/book/operations/upgrading.md

@@ -71,7 +71,7 @@ The entire procedure would look something like this:
 
 2. Take each node offline individually to back up its metadata folder, bring them back online once the backup is done.
   You can do all of the nodes in a single zone at once as that won't impact global cluster availability.
-  Do not try to make a backup of the metadata folder of a running node.
+  Do not try to manually copy the metadata folder of a running node.
 
   **Since Garage v0.9.4,** you can use the `garage meta snapshot --all` command
   to take a simultaneous snapshot of the metadata database files of all your

doc/book/quick-start/_index.md

@@ -42,6 +42,13 @@ If a binary of the last version is not available for your architecture,
 or if you want a build customized for your system,
 you can [build Garage from source](@/documentation/cookbook/from-source.md).
 
+If none of these option work for you, you can also run Garage in a Docker
+container.  When using Docker, the commands used in this guide will not work
+anymore.  We recommend reading the tutorial on [configuring a
+multi-node cluster](@/documentation/cookbook/real-world.md) to learn about
+using Garage as a Docker container. For simplicity, a minimal command to launch
+Garage using Docker is provided in this quick start guide as well.
+
 
 ## Configuring and starting Garage
 
@@ -85,6 +92,9 @@ metrics_token = "$(openssl rand -base64 32)"
 EOF
 ```
 
+See the [Configuration file format](https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/)
+for complete options and values.
+
 Now that your configuration file has been created, you may save it to the directory of your choice.
 By default, Garage looks for **`/etc/garage.toml`.**
 You can also store it somewhere else, but you will have to specify `-c path/to/garage.toml`
@@ -111,6 +121,26 @@ garage -c path/to/garage.toml server
 
 If you have placed the `garage.toml` file in `/etc` (its default location), you can simply run `garage server`.
 
+Alternatively, if you cannot or do not wish to run the Garage binary directly,
+you may use Docker to run Garage in a container using the following command:
+
+```bash
+docker run \
+  -d \
+  --name garaged \
+  -p 3900:3900 -p 3901:3901 -p 3902:3902 -p 3903:3903 \
+  -v /path/to/garage.toml:/etc/garage.toml \
+  -v /path/to/garage/meta:/var/lib/garage/meta \
+  -v /path/to/garage/data:/var/lib/garage/data \
+  dxflrs/garage:v1.3.0
+```
+
+Under Linux, you can substitute `--network host` for `-p 3900:3900 -p 3901:3901 -p 3902:3902 -p 3903:3903`
+
+#### Troubleshooting
+
+Ensure your configuration file, `metadata_dir` and `data_dir`  are readable by the user running the `garage` server or Docker.
+
 You can tune Garage's verbosity by setting the `RUST_LOG=` environment variable. \
 Available log levels are (from less verbose to more verbose): `error`, `warn`, `info` *(default)*, `debug` and `trace`.
 
@@ -131,6 +161,9 @@ It uses values from the TOML configuration file to find the Garage daemon runnin
 local node, therefore if your configuration file is not at `/etc/garage.toml` you will
 again have to specify `-c path/to/garage.toml` at each invocation.
 
+If you are running Garage in a Docker container, you can set `alias garage="docker exec -ti <container name> /garage"`
+to use the Garage binary inside your container.
+
 If the `garage` CLI is able to correctly detect the parameters of your local Garage node,
 the following command should be enough to show the status of your cluster:
 
@@ -149,11 +182,12 @@ ID                 Hostname  Address         Tag                   Zone  Capacit
 ## Creating a cluster layout
 
 Creating a cluster layout for a Garage deployment means informing Garage
-of the disk space available on each node of the cluster
+of the disk space available on each node of the cluster, `-c`,
-as well as the zone (e.g. datacenter) each machine is located in.
+as well as the name of the zone (e.g. datacenter), `-z`, each machine is located in.
 
-For our test deployment, we are using only one node. The way in which we configure
+For our test deployment, we are have only one node with zone named `dc1` and a
-it does not matter, you can simply write:
+capacity of `1G`, though the capacity is ignored for a single node deployment
+and can be changed later when adding new nodes.
 
 ```bash
 garage layout assign -z dc1 -c 1G <node_id>
@@ -166,7 +200,7 @@ For instance here you could write just `garage layout assign -z dc1 -c 1G 563e`.
 The layout then has to be applied to the cluster, using:
 
 ```bash
-garage layout apply
+garage layout apply --version 1
 ```
 
 
@@ -316,7 +350,7 @@ Check [our s3 compatibility list](@/documentation/reference-manual/s3-compatibil
 
 ### Other tools for interacting with Garage
 
-The following tools can also be used to send and recieve files from/to Garage:
+The following tools can also be used to send and receive files from/to Garage:
 
 - [minio-client](@/documentation/connect/cli.md#minio-client) 
 - [s3cmd](@/documentation/connect/cli.md#s3cmd) 

doc/book/reference-manual/configuration.md

@@ -13,16 +13,19 @@ consistency_mode = "consistent"
 
 metadata_dir = "/var/lib/garage/meta"
 data_dir = "/var/lib/garage/data"
+metadata_snapshots_dir = "/var/lib/garage/snapshots"
 metadata_fsync = true
 data_fsync = false
 disable_scrub = false
+use_local_tz = false
 metadata_auto_snapshot_interval = "6h"
 
 db_engine = "lmdb"
 
 block_size = "1M"
 block_ram_buffer_max = "256MiB"
-
+block_max_concurrent_reads = 16
+block_max_concurrent_writes_per_request =10
 lmdb_map_size = "1T"
 
 compression_level = 1
@@ -44,6 +47,7 @@ bootstrap_peers = [
     "212fd62eeaca72c122b45a7f4fa0f55e012aa5e24ac384a72a3016413fa724ff@[fc00:F::1]:3901",
 ]
 
+allow_punycode = false
 
 [consul_discovery]
 api = "catalog"
@@ -73,6 +77,7 @@ root_domain = ".s3.garage"
 [s3_web]
 bind_addr = "[::]:3902"
 root_domain = ".web.garage"
+add_host_to_metrics = true
 
 [admin]
 api_bind_addr = "0.0.0.0:3903"
@@ -89,12 +94,16 @@ The following gives details about each available configuration option.
 
 [Environment variables](#env_variables).
 
-Top-level configuration options:
+Top-level configuration options, in alphabetical order:
+[`allow_punycode`](#allow_punycode),
 [`allow_world_readable_secrets`](#allow_world_readable_secrets),
+[`block_max_concurrent_reads`](`block_max_concurrent_reads),
 [`block_ram_buffer_max`](#block_ram_buffer_max),
+[`block_max_concurrent_writes_per_request`](#block_max_concurrent_writes_per_request),
 [`block_size`](#block_size),
 [`bootstrap_peers`](#bootstrap_peers),
 [`compression_level`](#compression_level),
+[`consistency_mode`](#consistency_mode),
 [`data_dir`](#data_dir),
 [`data_fsync`](#data_fsync),
 [`db_engine`](#db_engine),
@@ -103,13 +112,14 @@ Top-level configuration options:
 [`metadata_auto_snapshot_interval`](#metadata_auto_snapshot_interval),
 [`metadata_dir`](#metadata_dir),
 [`metadata_fsync`](#metadata_fsync),
+[`metadata_snapshots_dir`](#metadata_snapshots_dir),
 [`replication_factor`](#replication_factor),
-[`consistency_mode`](#consistency_mode),
 [`rpc_bind_addr`](#rpc_bind_addr),
 [`rpc_bind_outgoing`](#rpc_bind_outgoing),
 [`rpc_public_addr`](#rpc_public_addr),
 [`rpc_public_addr_subnet`](#rpc_public_addr_subnet)
-[`rpc_secret`/`rpc_secret_file`](#rpc_secret).
+[`rpc_secret`/`rpc_secret_file`](#rpc_secret),
+[`use_local_tz`](#use_local_tz).
 
 The `[consul_discovery]` section:
 [`api`](#consul_api),
@@ -134,6 +144,7 @@ The `[s3_api]` section:
 [`s3_region`](#s3_region).
 
 The `[s3_web]` section:
+[`add_host_to_metrics`](#web_add_host_to_metrics),
 [`bind_addr`](#web_bind_addr),
 [`root_domain`](#web_root_domain).
 
@@ -145,13 +156,17 @@ The `[admin]` section:
 
 ### Environment variables {#env_variables}
 
-The following configuration parameter must be specified as an environment
+The following configuration parameters must be specified as environment variables,
-variable, it does not exist in the configuration file:
+they do not exist in the configuration file:
 
-- `GARAGE_LOG_TO_SYSLOG` (since v0.9.4): set this to `1` or `true` to make the
+- `GARAGE_LOG_TO_SYSLOG` (since `v0.9.4`): set this to `1` or `true` to make the
   Garage daemon send its logs to `syslog` (using the libc `syslog` function)
   instead of printing to stderr.
 
+- `GARAGE_LOG_TO_JOURNALD` (since `v1.2.0`): set this to `1` or `true` to make the
+  Garage daemon send its logs to `journald` (using the native protocol of `systemd-journald`)
+  instead of printing to stderr.
+
 The following environment variables can be used to override the corresponding
 values in the configuration file:
 
@@ -163,7 +178,7 @@ values in the configuration file:
 
 ### Top-level configuration options
 
-#### `replication_factor` {#replication_factor}
+#### `replication_factor` (since `v1.0.0`) {#replication_factor}
 
 The replication factor can be any positive integer smaller or equal the node count in your cluster.
 The chosen replication factor has a big impact on the cluster's failure tolerancy and performance characteristics.
@@ -211,7 +226,7 @@ is in progress.  In theory, no data should be lost as rebalancing is a
 routine operation for Garage, although we cannot guarantee you that everything
  will go right in such an extreme scenario.
 
-#### `consistency_mode` {#consistency_mode}
+#### `consistency_mode` (since `v1.0.0`) {#consistency_mode}
 
 The consistency mode setting determines the read and write behaviour of your cluster.
 
@@ -273,6 +288,7 @@ as the index of all objects, object version and object blocks.
 
 Store this folder on a fast SSD drive if possible to maximize Garage's performance.
 
+
 #### `data_dir` {#data_dir}
 
 The directory in which Garage will store the data blocks of objects.
@@ -293,6 +309,25 @@ data_dir = [
 See [the dedicated documentation page](@/documentation/operations/multi-hdd.md)
 on how to operate Garage in such a setup.
 
+#### `metadata_snapshots_dir` (since `v1.1.0`) {#metadata_snapshots_dir}
+
+The directory in which Garage will store metadata snapshots when it
+performs a snapshot of the metadata database, either when instructed to do
+so from a RPC call or regularly through
+[`metadata_auto_snapshot_interval`](#metadata_auto_snapshot_interval).
+
+By default, Garage will store snapshots into a `snapshots/` subdirectory
+of [`metadata_dir`](#metadata_dir). This might quickly fill up your
+metadata storage space if you use snapshots, because Garage will need up
+to 4x the space of the existing metadata database: each snapshot requires
+roughly as much space as the original database, and Garage temporarily
+needs to store up to three different snapshots before it cleans up the oldest
+snapshot to go back to two stored snapshots.
+
+To prevent filling your disk, you might to change this setting to a
+directory with ample available space, e.g. on the same storage space as
+[`data_dir`](#data_dir).
+
 #### `db_engine` (since `v0.8.0`) {#db_engine}
 
 Since `v0.8.0`, Garage can use alternative storage backends as follows:
@@ -301,6 +336,7 @@ Since `v0.8.0`, Garage can use alternative storage backends as follows:
 | --------- | ----------------- | ------------- |
 | [LMDB](https://www.symas.com/lmdb) (since `v0.8.0`, default since `v0.9.0`) | `"lmdb"` | `<metadata_dir>/db.lmdb/` |
 | [Sqlite](https://sqlite.org) (since `v0.8.0`) | `"sqlite"` | `<metadata_dir>/db.sqlite` |
+| [Fjall](https://github.com/fjall-rs/fjall) (**experimental support** since `v1.3.0`) | `"fjall"` | `<metadata_dir>/db.fjall/` |
 | [Sled](https://sled.rs) (old default, removed since `v1.0`) | `"sled"` | `<metadata_dir>/db/` |
 
 Sled was supported until Garage v0.9.x, and was removed in Garage v1.0.
@@ -337,6 +373,14 @@ LMDB works very well, but is known to have the following limitations:
   so it is not the best choice for high-performance storage clusters,
   but it should work fine in many cases.
 
+- Fjall: a storage engine based on LSM trees, which theoretically allow for
+  higher write throughput than other storage engines that are based on B-trees.
+  Using Fjall could potentially improve Garage's performance significantly in
+  write-heavy workloads. **Support for Fjall is experimental at this point**,
+  we have added it to Garage for evaluation purposes only. **Do not use it for
+  production-critical workloads.**
+
+
 It is possible to convert Garage's metadata directory from one format to another
 using the `garage convert-db` command, which should be used as follows:
 
@@ -374,6 +418,7 @@ Here is how this option impacts the different database engines:
 |----------|------------------------------------|-------------------------------|
 | Sqlite   | `PRAGMA synchronous = OFF`         | `PRAGMA synchronous = NORMAL` |
 | LMDB     | `MDB_NOMETASYNC` + `MDB_NOSYNC`    | `MDB_NOMETASYNC`              |
+| Fjall    | default options                    | not supported                 |
 
 Note that the Sqlite database is always ran in `WAL` mode (`PRAGMA journal_mode = WAL`).
 
@@ -390,7 +435,7 @@ at the cost of a moderate drop in write performance.
 Similarly to `metatada_fsync`, this is likely not necessary
 if geographical replication is used.
 
-#### `metadata_auto_snapshot_interval` (since Garage v0.9.4) {#metadata_auto_snapshot_interval}
+#### `metadata_auto_snapshot_interval` (since `v0.9.4`) {#metadata_auto_snapshot_interval}
 
 If this value is set, Garage will automatically take a snapshot of the metadata
 DB file at a regular interval and save it in the metadata directory.
@@ -427,6 +472,13 @@ you should delete it from the data directory and then call `garage repair
 blocks` on the node to ensure that it re-obtains a copy from another node on
 the network.
 
+#### `use_local_tz` (since `v1.1.0`) {#use_local_tz}
+
+By default, Garage runs the lifecycle worker every day at midnight in UTC. Set the
+`use_local_tz` configuration value to `true` if you want Garage to run the
+lifecycle worker at midnight in your local timezone. If you have multiple nodes,
+you should also ensure that each node has the same timezone configuration.
+
 #### `block_size` {#block_size}
 
 Garage splits stored objects in consecutive chunks of size `block_size`
@@ -442,7 +494,7 @@ files will remain available. This however means that chunks from existing files
 will not be deduplicated with chunks from newly uploaded files, meaning you
 might use more storage space that is optimally possible.
 
-#### `block_ram_buffer_max` (since v0.9.4) {#block_ram_buffer_max}
+#### `block_ram_buffer_max` (since `v0.9.4`) {#block_ram_buffer_max}
 
 A limit on the total size of data blocks kept in RAM by S3 API nodes awaiting
 to be sent to storage nodes asynchronously.
@@ -473,6 +525,37 @@ node.
 
 The default value is 256MiB.
 
+#### `block_max_concurrent_reads` (since `v1.3.0` / `v2.1.0`) {#block_max_concurrent_reads}
+
+The maximum number of blocks (individual files in the data directory) open
+simultaneously for reading.
+
+Reducing this number does not limit the number of data blocks that can be
+transferred through the network simultaneously. This mechanism was just added
+as a backpressure mechanism for HDD read speed: it helps avoid a situation
+where too many requests are coming in and Garage is reading too many block
+files simultaneously, thus not making timely progress on any of the reads.
+
+When a request to read a data block comes in through the network, the requests
+awaits for one of the `block_max_concurrent_reads` slots to be available
+(internally implemented using a Semaphore object). Once it acquired a read
+slot, it reads the entire block file to RAM and frees the slot as soon as the
+block file is finished reading. Only after the slot is released will the
+block's data start being transferred over the network.  If the request fails to
+acquire a reading slot wihtin 15 seconds, it fails with a timeout error.
+Timeout events can be monitored through the `block_read_semaphore_timeouts`
+metric in Prometheus: a non-zero number of such events indicates an I/O
+bottleneck on HDD read speed.
+
+
+#### `block_max_concurrent_writes_per_request` (since  `v2.1.0`) {#block_max_concurrent_writes_per_request}
+
+This parameter is designed to adapt to the concurrent write performance of 
+different storage media.Maximum number of parallel block writes per put request 
+Higher values improve throughput but increase memory usage.
+
+Default: 3, Recommended: 10-30 for NVMe, 3-10 for HDD
+
 #### `lmdb_map_size` {#lmdb_map_size}
 
 This parameters can be used to set the map size used by LMDB,
@@ -529,7 +612,7 @@ the node, even in the case of a NAT: the NAT should be configured to forward the
 port number to the same internal port nubmer. This means that if you have several nodes running
 behind a NAT, they should each use a different RPC port number.
 
-#### `rpc_bind_outgoing`(since v0.9.2) {#rpc_bind_outgoing}
+#### `rpc_bind_outgoing` (since `v0.9.2`) {#rpc_bind_outgoing}
 
 If enabled, pre-bind all sockets for outgoing connections to the same IP address
 used for listening (the IP address specified in `rpc_bind_addr`) before
@@ -571,7 +654,7 @@ be obtained by running `garage node id` and then included directly in the
 key will be returned by `garage node id` and you will have to add the IP
 yourself.
 
-### `allow_world_readable_secrets` or `GARAGE_ALLOW_WORLD_READABLE_SECRETS` (env) {#allow_world_readable_secrets}
+#### `allow_world_readable_secrets` or `GARAGE_ALLOW_WORLD_READABLE_SECRETS` (env) {#allow_world_readable_secrets}
 
 Garage checks the permissions of your secret files to make sure they're not
 world-readable. In some cases, the check might fail and consider your files as
@@ -583,6 +666,13 @@ permission verification.
 Alternatively, you can set the `GARAGE_ALLOW_WORLD_READABLE_SECRETS`
 environment variable to `true` to bypass the permissions check.
 
+#### `allow_punycode` {#allow_punycode}
+
+Allow creating buckets with names containing punycode. When used for buckets served
+as websites, this allows using almost any unicode character in the domain name.
+
+Default to `false`.
+
 ### The `[consul_discovery]` section
 
 Garage supports discovering other nodes of the cluster using Consul.  For this
@@ -713,6 +803,13 @@ For instance, if `root_domain` is `web.garage.eu`, a bucket called `deuxfleurs.f
 will be accessible either with hostname `deuxfleurs.fr.web.garage.eu`
 or with hostname `deuxfleurs.fr`.
 
+#### `add_host_to_metrics` {#web_add_host_to_metrics}
+
+Whether to include the requested domain name (HTTP `Host` header) in the
+Prometheus metrics of the web endpoint. This is disabled by default as the
+number of possible values is not bounded and can be a source of cardinality
+explosion in the exported metrics.
+
 
 ### The `[admin]` section
 

doc/book/reference-manual/features.md

@@ -61,7 +61,7 @@ directed to a Garage cluster can be handled independently of one another instead
 of going through a central bottleneck (the leader node).
 As a consequence, requests can be handled much faster, even in cases where latency
 between cluster nodes is important (see our [benchmarks](@/documentation/design/benchmarks/index.md) for data on this).
-This is particularly usefull when nodes are far from one another and talk to one other through standard Internet connections.
+This is particularly useful when nodes are far from one another and talk to one other through standard Internet connections.
 
 ### Web server for static websites
 

doc/book/reference-manual/monitoring.md

@@ -392,7 +392,7 @@ table_merkle_updater_todo_queue_length{table_name="block_ref"} 0
 
 #### `table_sync_items_received`, `table_sync_items_sent` (counters)
 
-Number of data items sent to/recieved from other nodes during resync procedures
+Number of data items sent to/received from other nodes during resync procedures
 
 ```
 table_sync_items_received{from="<remote node>",table_name="bucket_v2"} 3

doc/book/reference-manual/s3-compatibility.md

@@ -23,17 +23,17 @@ Feel free to open a PR to suggest fixes this table. Minio is missing because the
 - 2022-05-25 - Many Ceph S3 endpoints are not documented but implemented. Following a notification from the Ceph community, we added them.
 
 
-
 ## High-level features
 
 | Feature                      | Garage                           | [Openstack Swift](https://docs.openstack.org/swift/latest/s3_compat.html) | [Ceph Object Gateway](https://docs.ceph.com/en/latest/radosgw/s3/) | [Riak CS](https://docs.riak.com/riak/cs/2.1.1/references/apis/storage/s3/index.html) | [OpenIO](https://docs.openio.io/latest/source/arch-design/s3_compliancy.html) |
 |------------------------------|----------------------------------|-----------------|---------------|---------|-----|
-| [signature v2](https://docs.aws.amazon.com/general/latest/gr/signature-version-2.html) (deprecated) | ❌ Missing | ✅ |  ✅ | ✅ | ✅ |
+| [signature v2](https://docs.aws.amazon.com/AmazonS3/latest/API/Appendix-Sigv2.html) (deprecated) | ❌ Missing | ✅ |  ✅ | ✅ | ✅ |
 | [signature v4](https://docs.aws.amazon.com/AmazonS3/latest/API/sig-v4-authenticating-requests.html) |  ✅ Implemented |  ✅ |  ✅ | ❌ | ✅ |
 | [URL path-style](https://docs.aws.amazon.com/AmazonS3/latest/userguide/VirtualHosting.html#path-style-access) (eg. `host.tld/bucket/key`) |  ✅ Implemented | ✅ |  ✅ | ❓| ✅ |
 | [URL vhost-style](https://docs.aws.amazon.com/AmazonS3/latest/userguide/VirtualHosting.html#virtual-hosted-style-access) URL (eg. `bucket.host.tld/key`) |  ✅ Implemented | ❌| ✅| ✅ | ✅ |
 | [Presigned URLs](https://docs.aws.amazon.com/AmazonS3/latest/userguide/ShareObjectPreSignedURL.html) |  ✅ Implemented | ❌|  ✅ | ✅ |  ✅(❓) |
 | [SSE-C encryption](https://docs.aws.amazon.com/AmazonS3/latest/userguide/ServerSideEncryptionCustomerKeys.html) |  ✅ Implemented | ❓ |  ✅ | ❌ |  ✅ |
+| [Bucket versioning](https://docs.aws.amazon.com/AmazonS3/latest/userguide/Versioning.html) | ❌ Missing | ✅ |  ✅ | ❌ | ✅ |
 
 *Note:* OpenIO does not says if it supports presigned URLs. Because it is part
 of signature v4 and they claim they support it without additional precisions,

doc/book/working-documents/design-draft.md

@@ -42,7 +42,7 @@ The general principle are similar, but details have not been updated.**
 A version is defined by the existence of at least one entry in the blocks table for a certain version UUID.
 We must keep the following invariant: if a version exists in the blocks table, it has to be referenced in the objects table.
 We explicitly manage concurrent versions of an object: the version timestamp and version UUID columns are index columns, thus we may have several concurrent versions of an object.
-Important: before deleting an older version from the objects table, we must make sure that we did a successfull delete of the blocks of that version from the blocks table.
+Important: before deleting an older version from the objects table, we must make sure that we did a successful delete of the blocks of that version from the blocks table.
 
 Thus, the workflow for reading an object is as follows:
 
@@ -95,7 +95,7 @@ Known issue: if someone is reading from a version that we want to delete and the
 Usefull metadata:
 
 - list of versions that reference this block in the Casandra table, so that we can do GC by checking in Cassandra that the lines still exist
-- list of other nodes that we know have acknowledged a write of this block, usefull in the rebalancing algorithm
+- list of other nodes that we know have acknowledged a write of this block, useful in the rebalancing algorithm
 
 Write strategy: have a single thread that does all write IO so that it is serialized (or have several threads that manage independent parts of the hash space). When writing a blob, write it to a temporary file, close, then rename so that a concurrent read gets a consistent result (either not found or found with whole content).
 

doc/book/working-documents/migration-04.md

@@ -68,7 +68,7 @@ The migration steps are as follows:
 5. Turn off Garage 0.3
 
 6. Backup metadata folders if you can (i.e. if you have space to do it
-   somewhere). Backuping data folders could also be usefull but that's much
+   somewhere). Backuping data folders could also be useful but that's much
    harder to do. If your filesystem supports snapshots, this could be a good
    time to use them.
 

doc/book/working-documents/testing-strategy.md

@@ -37,7 +37,7 @@ There are two reasons for this:
 
 Reminder: rules of simplicity, concerning changes to Garage's source code.
 Always question what we are doing.
-Never do anything just because it looks nice or because we "think" it might be usefull at some later point but without knowing precisely why/when.
+Never do anything just because it looks nice or because we "think" it might be useful at some later point but without knowing precisely why/when.
 Only do things that make perfect sense in the context of what we currently know.
 
 ## References

doc/drafts/admin-api.md

@@ -70,7 +70,7 @@ Example response body:
 ```json
 {
   "node": "b10c110e4e854e5aa3f4637681befac755154b20059ec163254ddbfae86b09df",
-  "garageVersion": "v1.0.0",
+  "garageVersion": "v1.3.0",
   "garageFeatures": [
     "k2v",
     "lmdb",

doc/drafts/k2v-spec.md

@@ -562,7 +562,7 @@ token>", v: ["<value1>", ...] }`, with the following fields:
 - in case of concurrent update and deletion, a `null` is added to the list of concurrent values
 
 - if the `tombstones` query parameter is set to `true`, tombstones are returned
-  for items that have been deleted (this can be usefull for inserting after an
+  for items that have been deleted (this can be useful for inserting after an
   item that has been deleted, so that the insert is not considered
   concurrent with the delete). Tombstones are returned as tuples in the
   same format with only `null` values

flake.lock

@@ -1,38 +1,27 @@
 {
   "nodes": {
-    "cargo2nix": {
+    "crane": {
-      "inputs": {
-        "flake-compat": [
-          "flake-compat"
-        ],
-        "flake-utils": "flake-utils",
-        "nixpkgs": [
-          "nixpkgs"
-        ],
-        "rust-overlay": "rust-overlay"
-      },
       "locked": {
-        "lastModified": 1666087781,
+        "lastModified": 1737689766,
-        "narHash": "sha256-trKVdjMZ8mNkGfLcY5LsJJGtdV3xJDZnMVrkFjErlcs=",
+        "narHash": "sha256-ivVXYaYlShxYoKfSo5+y5930qMKKJ8CLcAoIBPQfJ6s=",
-        "owner": "Alexis211",
+        "owner": "ipetkov",
-        "repo": "cargo2nix",
+        "repo": "crane",
-        "rev": "a7a61179b66054904ef6a195d8da736eaaa06c36",
+        "rev": "6fe74265bbb6d016d663b1091f015e2976c4a527",
         "type": "github"
       },
       "original": {
-        "owner": "Alexis211",
+        "owner": "ipetkov",
-        "repo": "cargo2nix",
+        "repo": "crane",
-        "rev": "a7a61179b66054904ef6a195d8da736eaaa06c36",
         "type": "github"
       }
     },
     "flake-compat": {
       "locked": {
-        "lastModified": 1688025799,
+        "lastModified": 1717312683,
-        "narHash": "sha256-ktpB4dRtnksm9F5WawoIkEneh1nrEvuxb5lJFt1iOyw=",
+        "narHash": "sha256-FrlieJH50AuvagamEvWMIE6D2OAnERuDboFDYAED/dE=",
         "owner": "nix-community",
         "repo": "flake-compat",
-        "rev": "8bf105319d44f6b9f0d764efa4fdef9f1cc9ba1c",
+        "rev": "38fd3954cf65ce6faf3d0d45cd26059e059f07ea",
         "type": "github"
       },
       "original": {
@@ -46,29 +35,11 @@
         "systems": "systems"
       },
       "locked": {
-        "lastModified": 1681202837,
+        "lastModified": 1731533236,
-        "narHash": "sha256-H+Rh19JDwRtpVPAWp64F+rlEtxUWBAQW28eAi3SRSzg=",
+        "narHash": "sha256-l0KFg5HjrsfsO/JpG+r7fRrqm12kzFHyUHqHCVpMMbI=",
         "owner": "numtide",
         "repo": "flake-utils",
-        "rev": "cfacdce06f30d2b68473a46042957675eebb3401",
+        "rev": "11707dc2f618dd54ca8739b309ec4fc024de578b",
-        "type": "github"
-      },
-      "original": {
-        "owner": "numtide",
-        "repo": "flake-utils",
-        "type": "github"
-      }
-    },
-    "flake-utils_2": {
-      "inputs": {
-        "systems": "systems_2"
-      },
-      "locked": {
-        "lastModified": 1681202837,
-        "narHash": "sha256-H+Rh19JDwRtpVPAWp64F+rlEtxUWBAQW28eAi3SRSzg=",
-        "owner": "numtide",
-        "repo": "flake-utils",
-        "rev": "cfacdce06f30d2b68473a46042957675eebb3401",
         "type": "github"
       },
       "original": {
@@ -79,63 +50,47 @@
     },
     "nixpkgs": {
       "locked": {
-        "lastModified": 1682109806,
+        "lastModified": 1763977559,
-        "narHash": "sha256-d9g7RKNShMLboTWwukM+RObDWWpHKaqTYXB48clBWXI=",
+        "narHash": "sha256-g4MKqsIRy5yJwEsI+fYODqLUnAqIY4kZai0nldAP6EM=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "2362848adf8def2866fabbffc50462e929d7fffb",
+        "rev": "cfe2c7d5b5d3032862254e68c37a6576b633d632",
-        "type": "github"
-      },
-      "original": {
-        "owner": "NixOS",
-        "ref": "nixpkgs-unstable",
-        "repo": "nixpkgs",
-        "type": "github"
-      }
-    },
-    "nixpkgs_2": {
-      "locked": {
-        "lastModified": 1707091808,
-        "narHash": "sha256-LahKBAfGbY836gtpVNnWwBTIzN7yf/uYM/S0g393r0Y=",
-        "owner": "NixOS",
-        "repo": "nixpkgs",
-        "rev": "9f2ee8c91ac42da3ae6c6a1d21555f283458247e",
         "type": "github"
       },
       "original": {
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "9f2ee8c91ac42da3ae6c6a1d21555f283458247e",
+        "rev": "cfe2c7d5b5d3032862254e68c37a6576b633d632",
         "type": "github"
       }
     },
     "root": {
       "inputs": {
-        "cargo2nix": "cargo2nix",
+        "crane": "crane",
         "flake-compat": "flake-compat",
-        "flake-utils": [
+        "flake-utils": "flake-utils",
-          "cargo2nix",
+        "nixpkgs": "nixpkgs",
-          "flake-utils"
+        "rust-overlay": "rust-overlay"
-        ],
-        "nixpkgs": "nixpkgs_2"
       }
     },
     "rust-overlay": {
       "inputs": {
-        "flake-utils": "flake-utils_2",
+        "nixpkgs": [
-        "nixpkgs": "nixpkgs"
+          "nixpkgs"
+        ]
       },
       "locked": {
-        "lastModified": 1707271822,
+        "lastModified": 1763952169,
-        "narHash": "sha256-/DZsoPH5GBzOpVEGz5PgJ7vh8Q6TcrJq5u8FcBjqAfI=",
+        "narHash": "sha256-+PeDBD8P+NKauH+w7eO/QWCIp8Cx4mCfWnh9sJmy9CM=",
         "owner": "oxalica",
         "repo": "rust-overlay",
-        "rev": "7a94fe7690d2bdfe1aab475382a505e14dc114a6",
+        "rev": "ab726555a9a72e6dc80649809147823a813fa95b",
         "type": "github"
       },
       "original": {
         "owner": "oxalica",
         "repo": "rust-overlay",
+        "rev": "ab726555a9a72e6dc80649809147823a813fa95b",
         "type": "github"
       }
     },
@@ -153,21 +108,6 @@
         "repo": "default",
         "type": "github"
       }
-    },
-    "systems_2": {
-      "locked": {
-        "lastModified": 1681028828,
-        "narHash": "sha256-Vy1rq5AaRuLzOxct8nz4T6wlgyUR7zLU309k9mBC768=",
-        "owner": "nix-systems",
-        "repo": "default",
-        "rev": "da67096a3b9bf56a91d16901293e51ba5b49a27e",
-        "type": "github"
-      },
-      "original": {
-        "owner": "nix-systems",
-        "repo": "default",
-        "type": "github"
-      }
     }
   },
   "root": "root",

flake.nix

@@ -2,89 +2,95 @@
   description =
     "Garage, an S3-compatible distributed object store for self-hosted deployments";
 
-  # Nixpkgs 23.11 as of 2024-02-07, has rustc v1.73
+  # Nixpkgs 25.05 as of 2025-11-24
   inputs.nixpkgs.url =
-    "github:NixOS/nixpkgs/9f2ee8c91ac42da3ae6c6a1d21555f283458247e";
+    "github:NixOS/nixpkgs/cfe2c7d5b5d3032862254e68c37a6576b633d632";
+
+  # Rust overlay as of 2025-11-24
+  inputs.rust-overlay.url =
+  "github:oxalica/rust-overlay/ab726555a9a72e6dc80649809147823a813fa95b";
+  inputs.rust-overlay.inputs.nixpkgs.follows = "nixpkgs";
+
+  inputs.crane.url = "github:ipetkov/crane";
 
   inputs.flake-compat.url = "github:nix-community/flake-compat";
+  inputs.flake-utils.url = "github:numtide/flake-utils";
 
-  inputs.cargo2nix = {
+  outputs = { self, nixpkgs, flake-utils, crane, rust-overlay, ... }:
-    # As of 2022-10-18: two small patches over unstable branch, one for clippy and one to fix feature detection
-    url = "github:Alexis211/cargo2nix/a7a61179b66054904ef6a195d8da736eaaa06c36";
-
-    # As of 2023-04-25:
-    # - my two patches were merged into unstable (one for clippy and one to "fix" feature detection)
-    # - rustc v1.66
-    # url = "github:cargo2nix/cargo2nix/8fb57a670f7993bfc24099c33eb9c5abb51f29a2";
-
-    # Rust overlay as of 2024-02-07
-    inputs.rust-overlay.url =
-      "github:oxalica/rust-overlay/7a94fe7690d2bdfe1aab475382a505e14dc114a6";
-
-    inputs.nixpkgs.follows = "nixpkgs";
-    inputs.flake-compat.follows = "flake-compat";
-  };
-
-  inputs.flake-utils.follows = "cargo2nix/flake-utils";
-
-  outputs = { self, nixpkgs, cargo2nix, flake-utils, ... }:
     let
-      git_version = self.lastModifiedDate;
       compile = import ./nix/compile.nix;
     in
     flake-utils.lib.eachDefaultSystem (system:
       let
         pkgs = nixpkgs.legacyPackages.${system};
+        packageFor = target: release: (compile {
+          inherit system target nixpkgs crane rust-overlay release;
+        }).garage;
+        testWith = extraTestEnv: (compile {
+          inherit system nixpkgs crane rust-overlay extraTestEnv;
+          release = false;
+        }).garage-test;
+        lints = (compile {
+          inherit system nixpkgs crane rust-overlay;
+          release = false;
+        });
       in
       {
-        packages =
+        packages = {
-          let
+          # default = native release build
-            packageFor = target: (compile {
+          default = packageFor null true;
-              inherit system git_version target;
+
-              pkgsSrc = nixpkgs;
+          # <arch> = cross-compiled, statically-linked release builds
-              cargo2nixOverlay = cargo2nix.overlays.default;
+          amd64 = packageFor "x86_64-unknown-linux-musl" true;
-              release = true;
+          i386 = packageFor "i686-unknown-linux-musl" true;
-            }).workspace.garage { compileMode = "build"; };
+          arm64 = packageFor "aarch64-unknown-linux-musl" true;
-          in
+          arm = packageFor "armv6l-unknown-linux-musl" true;
-          {
+
-            # default = native release build
+          # dev = native dev build
-            default = packageFor null;
+          dev = packageFor null false;
-            # other = cross-compiled, statically-linked builds
+
-            amd64 = packageFor "x86_64-unknown-linux-musl";
+          # test = cargo test
-            i386 = packageFor "i686-unknown-linux-musl";
+          tests = testWith {};
-            arm64 = packageFor "aarch64-unknown-linux-musl";
+          tests-lmdb = testWith {
-            arm = packageFor "armv6l-unknown-linux-musl";
+            GARAGE_TEST_INTEGRATION_DB_ENGINE = "lmdb";
           };
+          tests-sqlite = testWith {
+            GARAGE_TEST_INTEGRATION_DB_ENGINE = "sqlite";
+          };
+          tests-fjall = testWith {
+            GARAGE_TEST_INTEGRATION_DB_ENGINE = "fjall";
+          };
+
+          # lints (fmt, clippy)
+          fmt = lints.garage-cargo-fmt;
+          clippy = lints.garage-cargo-clippy;
+        };
 
         # ---- developpment shell, for making native builds only ----
         devShells =
           let
-            shellWithPackages = (packages: (compile {
+            targets = compile {
-              inherit system git_version;
+              inherit system nixpkgs crane rust-overlay;
-              pkgsSrc = nixpkgs;
+            };
-              cargo2nixOverlay = cargo2nix.overlays.default;
-            }).workspaceShell { inherit packages; });
           in
           {
-            default = shellWithPackages
+            default = targets.devShell;
-              (with pkgs; [
-                rustfmt
-                clang
-                mold
-              ]);
 
             # import the full shell using `nix develop .#full`
-            full = shellWithPackages (with pkgs; [
+            full = pkgs.mkShell {
-              rustfmt
+              buildInputs = with pkgs; [
-              rust-analyzer
+                targets.toolchain
-              clang
+                protobuf
-              mold
+                clang
-              # ---- extra packages for dev tasks ----
+                mold
-              cargo-audit
+                # ---- extra packages for dev tasks ----
-              cargo-outdated
+                rust-analyzer
-              cargo-machete
+                cargo-audit
-              nixpkgs-fmt
+                cargo-outdated
-            ]);
+                cargo-machete
+                nixpkgs-fmt
+              ];
+            };
           };
       });
 }

nix/build_index.nix

@@ -2,7 +2,7 @@
 
 with import ./common.nix;
 let
-  pkgs = import pkgsSrc { };
+  pkgs = import nixpkgs { };
   lib = pkgs.lib;
 
   /* Converts a key list and a value list to a set

nix/common.nix

@@ -10,9 +10,9 @@ let
 
   flake = (import flake-compat { system = builtins.currentSystem; src = ../.; });
 in
-rec {
+
-  pkgsSrc = flake.defaultNix.inputs.nixpkgs;
+{
-  cargo2nix = flake.defaultNix.inputs.cargo2nix;
+  flake = flake.defaultNix;
-  cargo2nixOverlay = cargo2nix.overlays.default;
+  nixpkgs = flake.defaultNix.inputs.nixpkgs;
-  devShells = builtins.getAttr builtins.currentSystem flake.defaultNix.devShells;
+  devShells = flake.defaultNix.devShells.${builtins.currentSystem};
 }

nix/compile.nix

@@ -1,164 +1,64 @@
-{ system, target ? null, pkgsSrc, cargo2nixOverlay, compiler ? "rustc"
+{
-, release ? false, git_version ? null, features ? null, }:
+  /* build inputs */
+  nixpkgs,
+  crane,
+  rust-overlay,
+
+  /* parameters */
+  system,
+  git_version ? null,
+  target ? null,
+  release ? false,
+  features ? null,
+  extraTestEnv ? {}
+}:
 
 let
   log = v: builtins.trace v v;
 
+  # NixOS and Rust/Cargo triples do not match for ARM, fix it here.
+  rustTarget = if target == "armv6l-unknown-linux-musleabihf" then
+    "arm-unknown-linux-musleabihf"
+  else
+    target;
+
+  rustTargetEnvMap = {
+    "x86_64-unknown-linux-musl" = "X86_64_UNKNOWN_LINUX_MUSL";
+    "aarch64-unknown-linux-musl" = "AARCH64_UNKNOWN_LINUX_MUSL";
+    "i686-unknown-linux-musl" = "I686_UNKNOWN_LINUX_MUSL";
+    "arm-unknown-linux-musleabihf" = "ARM_UNKNOWN_LINUX_MUSLEABIHF";
+  };
+
+  pkgsNative = import nixpkgs {
+    inherit system;
+    overlays = [ (import rust-overlay) ];
+  };
+
   pkgs = if target != null then
-    import pkgsSrc {
+    import nixpkgs {
       inherit system;
       crossSystem = {
         config = target;
         isStatic = true;
       };
-      overlays = [ cargo2nixOverlay ];
+      overlays = [ (import rust-overlay) ];
     }
   else
-    import pkgsSrc {
+    pkgsNative;
-      inherit system;
-      overlays = [ cargo2nixOverlay ];
-    };
 
-  toolchainOptions = {
+  inherit (pkgs) lib stdenv;
-    rustVersion = "1.73.0";
-    extraRustComponents = [ "clippy" ];
-  };
 
-  buildEnv = (drv:
+  toolchainFn = (p: p.rust-bin.stable."1.91.0".default.override {
-    {
+    targets = lib.optionals (target != null) [ rustTarget ];
-      rustc = drv.setBuildEnv;
+    extensions = [
-      clippy = ''
+      "rust-src"
-        ${drv.setBuildEnv or ""}
+      "rustfmt"
-        echo
-        echo --- BUILDING WITH CLIPPY ---
-        echo
-
-        export NIX_RUST_BUILD_FLAGS="''${NIX_RUST_BUILD_FLAGS} --deny warnings"
-        export RUSTC="''${CLIPPY_DRIVER}"
-      '';
-    }.${compiler});
-
-  /* Cargo2nix provides many overrides by default, you can take inspiration from them:
-     https://github.com/cargo2nix/cargo2nix/blob/master/overlay/overrides.nix
-
-     You can have a complete list of the available options by looking at the overriden object, mkcrate:
-     https://github.com/cargo2nix/cargo2nix/blob/master/overlay/mkcrate.nix
-  */
-  packageOverrides = pkgs:
-    pkgs.rustBuilder.overrides.all ++ [
-      /* [1] We add some logic to compile our crates with clippy, it provides us many additional lints
-
-         [2] We need to alter Nix hardening to make static binaries: PIE,
-         Position Independent Executables seems to be supported only on amd64. Having
-         this flag set either 1. make our executables crash or 2. compile as dynamic on some platforms.
-         Here, we deactivate it. Later (find `codegenOpts`), we reactivate it for supported targets
-         (only amd64 curently) through the `-static-pie` flag.
-         PIE is a feature used by ASLR, which helps mitigate security issues.
-         Learn more about Nix Hardening at: https://github.com/NixOS/nixpkgs/blob/master/pkgs/build-support/cc-wrapper/add-hardening.sh
-
-         [3] We want to inject the git version while keeping the build deterministic.
-         As we do not want to consider the .git folder as part of the input source,
-         we ask the user (the CI often) to pass the value to Nix.
-
-         [4] We don't want libsodium-sys and zstd-sys to try to use pkgconfig to build against a system library.
-         However the features to do so get activated for some reason (due to a bug in cargo2nix?),
-         so disable them manually here.
-      */
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "garage";
-        overrideAttrs = drv:
-          (if git_version != null then {
-            # [3]
-            preConfigure = ''
-              ${drv.preConfigure or ""}
-              export GIT_VERSION="${git_version}"
-            '';
-          } else
-            { }) // {
-              # [1]
-              setBuildEnv = (buildEnv drv);
-              # [2]
-              hardeningDisable = [ "pie" ];
-            };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "garage_rpc";
-        overrideAttrs = drv: { # [1]
-          setBuildEnv = (buildEnv drv);
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "garage_db";
-        overrideAttrs = drv: { # [1]
-          setBuildEnv = (buildEnv drv);
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "garage_util";
-        overrideAttrs = drv: { # [1]
-          setBuildEnv = (buildEnv drv);
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "garage_table";
-        overrideAttrs = drv: { # [1]
-          setBuildEnv = (buildEnv drv);
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "garage_block";
-        overrideAttrs = drv: { # [1]
-          setBuildEnv = (buildEnv drv);
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "garage_model";
-        overrideAttrs = drv: { # [1]
-          setBuildEnv = (buildEnv drv);
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "garage_api";
-        overrideAttrs = drv: { # [1]
-          setBuildEnv = (buildEnv drv);
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "garage_web";
-        overrideAttrs = drv: { # [1]
-          setBuildEnv = (buildEnv drv);
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "k2v-client";
-        overrideAttrs = drv: { # [1]
-          setBuildEnv = (buildEnv drv);
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "libsodium-sys";
-        overrideArgs = old: {
-          features = [ ]; # [4]
-        };
-      })
-
-      (pkgs.rustBuilder.rustLib.makeOverride {
-        name = "zstd-sys";
-        overrideArgs = old: {
-          features = [ ]; # [4]
-        };
-      })
     ];
+  });
+
+  craneLib = (crane.mkLib pkgs).overrideToolchain toolchainFn;
+
+  src = craneLib.cleanCargoSource ../.;
 
   /* We ship some parts of the code disabled by default by putting them behind a flag.
      It speeds up the compilation (when the feature is not required) and released crates have less dependency by default (less attack surface, disk space, etc.).
@@ -168,16 +68,16 @@ let
   rootFeatures = if features != null then
     features
   else
-    ([ "garage/bundled-libs" "garage/lmdb" "garage/sqlite" "garage/k2v" ] ++ (if release then [
+    ([ "bundled-libs" "lmdb" "sqlite" "fjall" "k2v" ] ++ (lib.optionals release [
-      "garage/consul-discovery"
+      "consul-discovery"
-      "garage/kubernetes-discovery"
+      "kubernetes-discovery"
-      "garage/metrics"
+      "metrics"
-      "garage/telemetry-otlp"
+      "telemetry-otlp"
-      "garage/syslog"
+      "syslog"
-    ] else
+      "journald"
-      [ ]));
+    ]));
 
-  packageFun = import ../Cargo.nix;
+  featuresStr = lib.concatStringsSep "," rootFeatures;
 
   /* We compile fully static binaries with musl to simplify deployment on most systems.
      When possible, we reactivate PIE hardening (see above).
@@ -188,12 +88,9 @@ let
      For more information on static builds, please refer to Rust's RFC 1721.
      https://rust-lang.github.io/rfcs/1721-crt-static.html#specifying-dynamicstatic-c-runtime-linkage
   */
-
+  codegenOptsMap = {
-  codegenOpts = {
+    "x86_64-unknown-linux-musl" =
-    "armv6l-unknown-linux-musleabihf" = [
+      [ "target-feature=+crt-static" "link-arg=-static-pie" ];
-      "target-feature=+crt-static"
-      "link-arg=-static"
-    ]; # compile as dynamic with static-pie
     "aarch64-unknown-linux-musl" = [
       "target-feature=+crt-static"
       "link-arg=-static"
@@ -202,17 +99,106 @@ let
       "target-feature=+crt-static"
       "link-arg=-static"
     ]; # segfault with static-pie
-    "x86_64-unknown-linux-musl" =
+    "armv6l-unknown-linux-musleabihf" = [
-      [ "target-feature=+crt-static" "link-arg=-static-pie" ];
+      "target-feature=+crt-static"
+      "link-arg=-static"
+    ]; # compile as dynamic with static-pie
   };
 
-  # NixOS and Rust/Cargo triples do not match for ARM, fix it here.
+  codegenOpts = if target != null then codegenOptsMap.${target} else [
-  rustTarget = if target == "armv6l-unknown-linux-musleabihf" then
+    "link-arg=-fuse-ld=mold"
-    "arm-unknown-linux-musleabihf"
+  ];
-  else
-    target;
 
-in pkgs.rustBuilder.makePackageSet ({
+  commonArgs =
-  inherit release packageFun packageOverrides codegenOpts rootFeatures;
+    {
-  target = rustTarget;
+      inherit src;
-} // toolchainOptions)
+      pname = "garage";
+      version = "dev";
+
+      strictDeps = true;
+      cargoExtraArgs = "--locked --features ${featuresStr}";
+      cargoTestExtraArgs = "--workspace";
+
+      nativeBuildInputs = [
+        pkgsNative.protobuf
+        pkgs.stdenv.cc
+      ] ++ lib.optionals (target == null) [
+        pkgs.clang
+        pkgs.mold
+      ];
+
+      CARGO_PROFILE = if release then "release" else "dev";
+      CARGO_BUILD_RUSTFLAGS =
+        lib.concatStringsSep
+          " "
+          (builtins.map (flag: "-C ${flag}") codegenOpts);
+    }
+  //
+    (if rustTarget != null then {
+      CARGO_BUILD_TARGET = rustTarget;
+
+      "CARGO_TARGET_${rustTargetEnvMap.${rustTarget}}_LINKER" = "${stdenv.cc.targetPrefix}cc";
+
+      HOST_CC = "${stdenv.cc.nativePrefix}cc";
+      TARGET_CC = "${stdenv.cc.targetPrefix}cc";
+    } else {
+      CARGO_TARGET_X86_64_UNKNOWN_LINUX_GNU_LINKER = "clang";
+    });
+
+in rec {
+  toolchain = toolchainFn pkgs;
+
+  devShell = pkgs.mkShell {
+    buildInputs = [
+      toolchain
+    ] ++ (with pkgs; [
+      protobuf
+      clang
+      mold
+    ]);
+  };
+
+  # ---- building garage ----
+
+  garage-deps = craneLib.buildDepsOnly commonArgs;
+
+  garage = craneLib.buildPackage (commonArgs // {
+    cargoArtifacts = garage-deps;
+
+    doCheck = false;
+  } //
+    (if git_version != null then {
+      version = git_version;
+      GIT_VERSION = git_version;
+    } else {}));
+
+  # ---- testing garage ----
+
+  garage-test-bin = craneLib.cargoBuild (commonArgs // {
+    cargoArtifacts = garage-deps;
+
+    pname = "garage-tests";
+
+    CARGO_PROFILE = "test";
+    cargoExtraArgs = "${commonArgs.cargoExtraArgs} --tests --workspace";
+    doCheck = false;
+  });
+
+  garage-test = craneLib.cargoTest (commonArgs // {
+    cargoArtifacts = garage-test-bin;
+    nativeBuildInputs = commonArgs.nativeBuildInputs ++ [
+      pkgs.cacert
+    ];
+  } // extraTestEnv);
+
+  # ---- source code linting ----
+
+  garage-cargo-fmt = craneLib.cargoFmt (commonArgs // {
+    cargoExtraArgs = "";
+  });
+
+  garage-cargo-clippy = craneLib.cargoClippy (commonArgs // {
+    cargoArtifacts = garage-deps;
+    cargoClippyExtraArgs = "--all-targets -- -D warnings";
+  });
+}

script/dev-cluster.sh

@@ -11,7 +11,7 @@ PATH="${GARAGE_DEBUG}:${GARAGE_RELEASE}:${NIX_RELEASE}:$PATH"
 FANCYCOLORS=("41m" "42m" "44m" "45m" "100m" "104m")
 
 export RUST_BACKTRACE=1 
-export RUST_LOG=garage=info,garage_api=debug
+export RUST_LOG=garage=info,garage_api_common=debug,garage_api_s3=debug
 MAIN_LABEL="\e[${FANCYCOLORS[0]}[main]\e[49m"
 
 if [ -z "$GARAGE_BIN" ]; then

script/dev-env-aws.sh

@@ -1,6 +1,7 @@
 export AWS_ACCESS_KEY_ID=`cat /tmp/garage.s3 |cut -d' ' -f1`
 export AWS_SECRET_ACCESS_KEY=`cat /tmp/garage.s3 |cut -d' ' -f2`
 export AWS_DEFAULT_REGION='garage'
+export AWS_REQUEST_CHECKSUM_CALCULATION='when_required'
 # FUTUREWORK: set AWS_ENDPOINT_URL instead, once nixpkgs bumps awscli to >=2.13.0.
 function aws { command aws --endpoint-url http://127.0.0.1:3911 $@ ; }
 

script/helm/README.md

@@ -1,3 +1,3 @@
 # Garage helm3 chart
 
-Documentation is located [here](/doc/book/cookbook/kubernetes.md).
+Documentation is located [here](https://garagehq.deuxfleurs.fr/documentation/cookbook/kubernetes/).

script/helm/garage/Chart.yaml

@@ -1,24 +1,18 @@
 apiVersion: v2
 name: garage
 description: S3-compatible object store for small self-hosted geo-distributed deployments
-
-# A chart can be either an 'application' or a 'library' chart.
-#
-# Application charts are a collection of templates that can be packaged into versioned archives
-# to be deployed.
-#
-# Library charts provide useful utilities or functions for the chart developer. They're included as
-# a dependency of application charts to inject those utilities and functions into the rendering
-# pipeline. Library charts do not define any templates and therefore cannot be deployed.
 type: application
+version: 0.7.3
+appVersion: "v1.3.1"
+home: https://garagehq.deuxfleurs.fr/
+icon: https://garagehq.deuxfleurs.fr/images/garage-logo.svg
 
-# This is the chart version. This version number should be incremented each time you make changes
+keywords:
-# to the chart and its templates, including the app version.
+- geo-distributed
-# Versions are expected to follow Semantic Versioning (https://semver.org/)
+- read-after-write-consistency
-version: 0.5.0
+- s3-compatible
 
-# This is the version number of the application being deployed. This version number should be
+sources:
-# incremented each time you make changes to the application. Versions are not expected to
+- https://git.deuxfleurs.fr/Deuxfleurs/garage.git
-# follow Semantic Versioning. They should reflect the version the application is using.
+
-# It is recommended to use it with quotes.
+maintainers: []
-appVersion: "v1.0.0"

script/helm/garage/README.md

@@ -0,0 +1,95 @@
+# garage
+
+![Version: 0.7.3](https://img.shields.io/badge/Version-0.7.3-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square) ![AppVersion: v1.3.1](https://img.shields.io/badge/AppVersion-v1.3.1-informational?style=flat-square)
+
+S3-compatible object store for small self-hosted geo-distributed deployments
+
+**Homepage:** <https://garagehq.deuxfleurs.fr/>
+
+## Source Code
+
+* <https://git.deuxfleurs.fr/Deuxfleurs/garage.git>
+
+## Values
+
+| Key | Type | Default | Description |
+|-----|------|---------|-------------|
+| affinity | object | `{}` |  |
+| deployment.kind | string | `"StatefulSet"` | Switchable to DaemonSet |
+| deployment.podManagementPolicy | string | `"OrderedReady"` | If using statefulset, allow Parallel or OrderedReady (default) |
+| deployment.replicaCount | int | `3` | Number of StatefulSet replicas/garage nodes to start |
+| environment | object | `{}` |  |
+| extraVolumeMounts | object | `{}` |  |
+| extraVolumes | object | `{}` |  |
+| fullnameOverride | string | `""` |  |
+| garage.blockSize | string | `"1048576"` | Defaults is 1MB An increase can result in better performance in certain scenarios https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#block-size |
+| garage.bootstrapPeers | list | `[]` | This is not required if you use the integrated kubernetes discovery |
+| garage.compressionLevel | string | `"1"` | zstd compression level of stored blocks https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#compression-level |
+| garage.dbEngine | string | `"lmdb"` | Can be changed for better performance on certain systems https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#db-engine-since-v0-8-0 |
+| garage.existingConfigMap | string | `""` | if not empty string, allow using an existing ConfigMap for the garage.toml, if set, ignores garage.toml |
+| garage.garageTomlString | string | `""` | String Template for the garage configuration if set, ignores above values. Values can be templated, see https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/ |
+| garage.kubernetesSkipCrd | bool | `false` | Set to true if you want to use k8s discovery but install the CRDs manually outside of the helm chart, for example if you operate at namespace level without cluster ressources |
+| garage.metadataAutoSnapshotInterval | string | `""` | If this value is set, Garage will automatically take a snapshot of the metadata DB file at a regular interval and save it in the metadata directory. https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#metadata_auto_snapshot_interval |
+| garage.replicationMode | string | `"3"` | Default to 3 replicas, see the replication_mode section at https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#replication-mode |
+| garage.rpcBindAddr | string | `"[::]:3901"` |  |
+| garage.rpcSecret | string | `""` | If not given, a random secret will be generated and stored in a Secret object |
+| garage.s3.api.region | string | `"garage"` |  |
+| garage.s3.api.rootDomain | string | `".s3.garage.tld"` |  |
+| garage.s3.web.index | string | `"index.html"` |  |
+| garage.s3.web.rootDomain | string | `".web.garage.tld"` |  |
+| image.pullPolicy | string | `"IfNotPresent"` |  |
+| image.repository | string | `"dxflrs/amd64_garage"` | default to amd64 docker image |
+| image.tag | string | `""` | set the image tag, please prefer using the chart version and not this to avoid compatibility issues |
+| imagePullSecrets | list | `[]` | set if you need credentials to pull your custom image |
+| ingress.s3.api.annotations | object | `{}` | Rely _either_ on the className or the annotation below but not both! If you want to use the className, set className: "nginx" and replace "nginx" by an Ingress controller name, examples [here](https://kubernetes.io/docs/concepts/services-networking/ingress-controllers). |
+| ingress.s3.api.enabled | bool | `false` |  |
+| ingress.s3.api.hosts[0] | object | `{"host":"s3.garage.tld","paths":[{"path":"/","pathType":"Prefix"}]}` | garage S3 API endpoint, to be used with awscli for example |
+| ingress.s3.api.hosts[1] | object | `{"host":"*.s3.garage.tld","paths":[{"path":"/","pathType":"Prefix"}]}` | garage S3 API endpoint, DNS style bucket access |
+| ingress.s3.api.labels | object | `{}` |  |
+| ingress.s3.api.tls | list | `[]` |  |
+| ingress.s3.web.annotations | object | `{}` | Rely _either_ on the className or the annotation below but not both! If you want to use the className, set className: "nginx" and replace "nginx" by an Ingress controller name, examples [here](https://kubernetes.io/docs/concepts/services-networking/ingress-controllers). |
+| ingress.s3.web.enabled | bool | `false` |  |
+| ingress.s3.web.hosts[0] | object | `{"host":"*.web.garage.tld","paths":[{"path":"/","pathType":"Prefix"}]}` | wildcard website access with bucket name prefix |
+| ingress.s3.web.hosts[1] | object | `{"host":"mywebpage.example.com","paths":[{"path":"/","pathType":"Prefix"}]}` | specific bucket access with FQDN bucket |
+| ingress.s3.web.labels | object | `{}` |  |
+| ingress.s3.web.tls | list | `[]` |  |
+| initImage.pullPolicy | string | `"IfNotPresent"` |  |
+| initImage.repository | string | `"busybox"` |  |
+| initImage.tag | string | `"stable"` |  |
+| livenessProbe | object | `{}` | Specifies a livenessProbe |
+| monitoring.metrics.enabled | bool | `false` | If true, a service for monitoring is created with a prometheus.io/scrape annotation |
+| monitoring.metrics.serviceMonitor.enabled | bool | `false` | If true, a ServiceMonitor CRD is created for a prometheus operator https://github.com/coreos/prometheus-operator |
+| monitoring.metrics.serviceMonitor.interval | string | `"15s"` |  |
+| monitoring.metrics.serviceMonitor.labels | object | `{}` |  |
+| monitoring.metrics.serviceMonitor.path | string | `"/metrics"` |  |
+| monitoring.metrics.serviceMonitor.relabelings | list | `[]` |  |
+| monitoring.metrics.serviceMonitor.scheme | string | `"http"` |  |
+| monitoring.metrics.serviceMonitor.scrapeTimeout | string | `"10s"` |  |
+| monitoring.metrics.serviceMonitor.tlsConfig | object | `{}` |  |
+| monitoring.tracing.sink | string | `""` | specify a sink endpoint for OpenTelemetry Traces, eg. `http://localhost:4317` |
+| nameOverride | string | `""` |  |
+| nodeSelector | object | `{}` |  |
+| persistence.data.hostPath | string | `"/var/lib/garage/data"` |  |
+| persistence.data.size | string | `"100Mi"` |  |
+| persistence.enabled | bool | `true` |  |
+| persistence.meta.hostPath | string | `"/var/lib/garage/meta"` |  |
+| persistence.meta.size | string | `"100Mi"` |  |
+| podAnnotations | object | `{}` | additonal pod annotations |
+| podSecurityContext.fsGroup | int | `1000` |  |
+| podSecurityContext.runAsGroup | int | `1000` |  |
+| podSecurityContext.runAsNonRoot | bool | `true` |  |
+| podSecurityContext.runAsUser | int | `1000` |  |
+| readinessProbe | object | `{}` | Specifies a readinessProbe |
+| resources | object | `{}` |  |
+| securityContext.capabilities | object | `{"drop":["ALL"]}` | The default security context is heavily restricted, feel free to tune it to your requirements |
+| securityContext.readOnlyRootFilesystem | bool | `true` |  |
+| service.s3.api.port | int | `3900` |  |
+| service.s3.web.port | int | `3902` |  |
+| service.type | string | `"ClusterIP"` | You can rely on any service to expose your cluster - ClusterIP (+ Ingress) - NodePort (+ Ingress) - LoadBalancer |
+| serviceAccount.annotations | object | `{}` | Annotations to add to the service account |
+| serviceAccount.create | bool | `true` | Specifies whether a service account should be created |
+| serviceAccount.name | string | `""` | The name of the service account to use. If not set and create is true, a name is generated using the fullname template |
+| tolerations | list | `[]` |  |
+
+----------------------------------------------
+Autogenerated from chart metadata using [helm-docs v1.14.2](https://github.com/norwoodj/helm-docs/releases/v1.14.2)

script/helm/garage/templates/configmap.yaml

@@ -1,7 +1,53 @@
+{{- if not .Values.garage.existingConfigMap }}
 apiVersion: v1
 kind: ConfigMap
 metadata:
   name: {{ include "garage.fullname" . }}-config
 data:
   garage.toml: |-
-    {{- tpl (index (index .Values.garage) "garage.toml") $ | nindent 4 }}
+    {{- if .Values.garage.garageTomlString }}
+        {{- tpl (index (index .Values.garage) "garageTomlString") $ | nindent 4 }}
+    {{- else }}
+    metadata_dir = "/mnt/meta"
+    data_dir = "/mnt/data"
+
+    db_engine = "{{ .Values.garage.dbEngine }}"
+
+    block_size = {{ .Values.garage.blockSize }}
+
+    replication_mode = "{{ .Values.garage.replicationMode }}"
+
+    compression_level = {{ .Values.garage.compressionLevel }}
+
+    {{- if .Values.garage.metadataAutoSnapshotInterval }}
+    metadata_auto_snapshot_interval = {{ .Values.garage.metadataAutoSnapshotInterval | quote }}
+    {{- end }}
+
+    rpc_bind_addr = "{{ .Values.garage.rpcBindAddr }}"
+    # rpc_secret will be populated by the init container from a k8s secret object
+    rpc_secret = "__RPC_SECRET_REPLACE__"
+
+    bootstrap_peers = {{ .Values.garage.bootstrapPeers }}
+
+    [kubernetes_discovery]
+    namespace = "{{ .Release.Namespace }}"
+    service_name = "{{ include "garage.fullname" . }}"
+    skip_crd = {{ .Values.garage.kubernetesSkipCrd }}
+
+    [s3_api]
+    s3_region = "{{ .Values.garage.s3.api.region }}"
+    api_bind_addr = "[::]:3900"
+    root_domain = "{{ .Values.garage.s3.api.rootDomain }}"
+
+    [s3_web]
+    bind_addr = "[::]:3902"
+    root_domain = "{{ .Values.garage.s3.web.rootDomain }}"
+    index = "{{ .Values.garage.s3.web.index }}"
+
+    [admin]
+    api_bind_addr = "[::]:3903"
+    {{- if .Values.monitoring.tracing.sink }}
+    trace_sink = "{{ .Values.monitoring.tracing.sink }}"
+    {{- end }}
+    {{- end }}
+{{- end }}

script/helm/garage/templates/service-headless.yaml

@@ -0,0 +1,22 @@
+{{- if eq .Values.deployment.kind "StatefulSet" -}}
+apiVersion: v1
+kind: Service
+metadata:
+  name: {{ include "garage.fullname" . }}-headless
+  labels:
+    {{- include "garage.labels" . | nindent 4 }}
+spec:
+  type: ClusterIP
+  clusterIP: None
+  ports:
+    - port: {{ .Values.service.s3.api.port }}
+      targetPort: 3900
+      protocol: TCP
+      name: s3-api
+    - port: {{ .Values.service.s3.web.port }}
+      targetPort: 3902
+      protocol: TCP
+      name: s3-web
+  selector:
+    {{- include "garage.selectorLabels" . | nindent 4 }}
+{{- end }}

script/helm/garage/templates/service.yaml

@@ -4,6 +4,10 @@ metadata:
   name: {{ include "garage.fullname" . }}
   labels:
     {{- include "garage.labels" . | nindent 4 }}
+  {{- with .Values.service.annotations }}
+  annotations:
+    {{- toYaml . | nindent 4 }}
+  {{- end }}
 spec:
   type: {{ .Values.service.type }}
   ports:
@@ -37,4 +41,4 @@ spec:
       name: metrics
   selector:
     {{- include "garage.selectorLabels" . | nindent 4 }}
-{{- end }}
+{{- end }}

script/helm/garage/templates/workload.yaml

@@ -10,12 +10,11 @@ spec:
       {{- include "garage.selectorLabels" . | nindent 6 }}
   {{- if eq .Values.deployment.kind "StatefulSet" }}
   replicas: {{ .Values.deployment.replicaCount }}
-  serviceName: {{ include "garage.fullname" . }}
+  serviceName: {{ include "garage.fullname" . }}-headless
   podManagementPolicy: {{ .Values.deployment.podManagementPolicy }}
   {{- end }}
   template:
     metadata:
-
       annotations:
         checksum/config: {{ include (print $.Template.BasePath "/configmap.yaml") . | sha256sum }}
       {{- with .Values.podAnnotations }}
@@ -76,15 +75,17 @@ spec:
             - name: etc
               mountPath: /etc/garage.toml
               subPath: garage.toml
-          # TODO
+          {{- with .Values.extraVolumeMounts }}
-          # livenessProbe:
+            {{- toYaml . | nindent 12 }}
-          #   httpGet:
+          {{- end }}
-          #     path: /
+          {{- with .Values.livenessProbe }}
-          #     port: 3900
+          livenessProbe:
-          # readinessProbe:
+            {{- toYaml . | nindent 12 }}
-          #   httpGet:
+          {{- end }}
-          #     path: /
+          {{- with .Values.readinessProbe }}
-          #     port: 3900
+          readinessProbe:
+            {{- toYaml . | nindent 12 }}
+          {{- end }}
           resources:
             {{- toYaml .Values.resources | nindent 12 }}
       volumes:
@@ -110,6 +111,9 @@ spec:
         - name: data
           emptyDir: {}
         {{- end }}
+        {{- with .Values.extraVolumes }}
+          {{- toYaml . | nindent 8 }}
+        {{- end }}
       {{- with .Values.nodeSelector }}
       nodeSelector:
         {{- toYaml . | nindent 8 }}

script/helm/garage/values.yaml

@@ -4,28 +4,34 @@
 
 # Garage configuration. These values go to garage.toml
 garage:
-  # Can be changed for better performance on certain systems
+  # -- Can be changed for better performance on certain systems
   # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#db-engine-since-v0-8-0
   dbEngine: "lmdb"
 
-  # Defaults is 1MB
+  # -- Defaults is 1MB
   # An increase can result in better performance in certain scenarios
   # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#block-size
   blockSize: "1048576"
 
-  # Default to 3 replicas, see the replication_mode section at
+  # -- Default to 3 replicas, see the replication_mode section at
   # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#replication-mode
   replicationMode: "3"
 
-  # zstd compression level of stored blocks
+  # -- zstd compression level of stored blocks
   # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#compression-level
   compressionLevel: "1"
 
+  # -- If this value is set, Garage will automatically take a snapshot of the metadata DB file at a regular interval and save it in the metadata directory.
+  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#metadata_auto_snapshot_interval
+  metadataAutoSnapshotInterval: ""
+
   rpcBindAddr: "[::]:3901"
-  # If not given, a random secret will be generated and stored in a Secret object
+  # -- If not given, a random secret will be generated and stored in a Secret object
   rpcSecret: ""
-  # This is not required if you use the integrated kubernetes discovery
+  # -- This is not required if you use the integrated kubernetes discovery
   bootstrapPeers: []
+  # -- Set to true if you want to use k8s discovery but install the CRDs manually outside
+  # of the helm chart, for example if you operate at namespace level without cluster ressources
   kubernetesSkipCrd: false
   s3:
     api:
@@ -34,47 +40,16 @@ garage:
     web:
       rootDomain: ".web.garage.tld"
       index: "index.html"
-  # Template for the garage configuration
-  # Values can be templated
-  # ref: https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/
-  garage.toml: |-
-    metadata_dir = "/mnt/meta"
-    data_dir = "/mnt/data"
 
-    db_engine = "{{ .Values.garage.dbEngine }}"
+  # -- if not empty string, allow using an existing ConfigMap for the garage.toml,
+  # if set, ignores garage.toml
+  existingConfigMap: ""
 
-    block_size = {{ .Values.garage.blockSize }}
+  # -- String Template for the garage configuration
-
+  # if set, ignores above values.
-    replication_mode = "{{ .Values.garage.replicationMode }}"
+  # Values can be templated,
-
+  # see https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/
-    compression_level = {{ .Values.garage.compressionLevel }}
+  garageTomlString: ""
-
-    rpc_bind_addr = "{{ .Values.garage.rpcBindAddr }}"
-    # rpc_secret will be populated by the init container from a k8s secret object
-    rpc_secret = "__RPC_SECRET_REPLACE__"
-
-    bootstrap_peers = {{ .Values.garage.bootstrapPeers }}
-
-    [kubernetes_discovery]
-    namespace = "{{ .Release.Namespace }}"
-    service_name = "{{ include "garage.fullname" . }}"
-    skip_crd = {{ .Values.garage.kubernetesSkipCrd }}
-
-    [s3_api]
-    s3_region = "{{ .Values.garage.s3.api.region }}"
-    api_bind_addr = "[::]:3900"
-    root_domain = "{{ .Values.garage.s3.api.rootDomain }}"
-
-    [s3_web]
-    bind_addr = "[::]:3902"
-    root_domain = "{{ .Values.garage.s3.web.rootDomain }}"
-    index = "{{ .Values.garage.s3.web.index }}"
-
-    [admin]
-    api_bind_addr = "[::]:3903"
-    {{- if .Values.monitoring.tracing.sink }}
-    trace_sink = "{{ .Values.monitoring.tracing.sink }}"
-    {{- end }}
 
 # Data persistence
 persistence:
@@ -92,16 +67,18 @@ persistence:
 
 # Deployment configuration
 deployment:
-  # Switchable to DaemonSet
+  # -- Switchable to DaemonSet
   kind: StatefulSet
-  # Number of StatefulSet replicas/garage nodes to start
+  # -- Number of StatefulSet replicas/garage nodes to start
   replicaCount: 3
-  # If using statefulset, allow Parallel or OrderedReady (default)
+  # -- If using statefulset, allow Parallel or OrderedReady (default)
   podManagementPolicy: OrderedReady
 
 image:
+  # -- default to amd64 docker image
   repository: dxflrs/amd64_garage
-  # please prefer using the chart version and not this tag
+  # -- set the image tag, please prefer using the chart version and not this
+  # to avoid compatibility issues
   tag: ""
   pullPolicy: IfNotPresent
 
@@ -110,19 +87,21 @@ initImage:
   tag: stable
   pullPolicy: IfNotPresent
 
+# -- set if you need credentials to pull your custom image
 imagePullSecrets: []
 nameOverride: ""
 fullnameOverride: ""
 
 serviceAccount:
-  # Specifies whether a service account should be created
+  # -- Specifies whether a service account should be created
   create: true
-  # Annotations to add to the service account
+  # -- Annotations to add to the service account
   annotations: {}
-  # The name of the service account to use.
+  # -- The name of the service account to use.
   # If not set and create is true, a name is generated using the fullname template
   name: ""
 
+# -- additonal pod annotations
 podAnnotations: {}
 
 podSecurityContext:
@@ -132,7 +111,7 @@ podSecurityContext:
   runAsNonRoot: true
 
 securityContext:
-  # The default security context is heavily restricted
+  # -- The default security context is heavily restricted,
   # feel free to tune it to your requirements
   capabilities:
     drop:
@@ -140,11 +119,13 @@ securityContext:
   readOnlyRootFilesystem: true
 
 service:
-  # You can rely on any service to expose your cluster
+  # -- You can rely on any service to expose your cluster
   # - ClusterIP (+ Ingress)
   # - NodePort (+ Ingress)
   # - LoadBalancer
   type: ClusterIP
+  # -- Annotations to add to the service
+  annotations: {}
   s3:
     api:
       port: 3900
@@ -156,20 +137,23 @@ ingress:
   s3:
     api:
       enabled: false
-      # Rely either on the className or the annotation below but not both
+      # -- Rely _either_ on the className or the annotation below but not both!
-      # replace "nginx" by an Ingress controller
+      # If you want to use the className, set
-      # you can find examples here https://kubernetes.io/docs/concepts/services-networking/ingress-controllers
       # className: "nginx"
+      # and replace "nginx" by an Ingress controller name,
+      # examples [here](https://kubernetes.io/docs/concepts/services-networking/ingress-controllers).
       annotations: {}
         # kubernetes.io/ingress.class: "nginx"
         # kubernetes.io/tls-acme: "true"
       labels: {}
       hosts:
-        - host: "s3.garage.tld" # garage S3 API endpoint
+        # -- garage S3 API endpoint, to be used with awscli for example
+        - host: "s3.garage.tld"
           paths:
             - path: /
               pathType: Prefix
-        - host: "*.s3.garage.tld" # garage S3 API endpoint, DNS style bucket access
+        # -- garage S3 API endpoint, DNS style bucket access
+        - host: "*.s3.garage.tld"
           paths:
             - path: /
               pathType: Prefix
@@ -179,20 +163,23 @@ ingress:
       #      - kubernetes.docker.internal
     web:
       enabled: false
-      # Rely either on the className or the annotation below but not both
+      # -- Rely _either_ on the className or the annotation below but not both!
-      # replace "nginx" by an Ingress controller
+      # If you want to use the className, set
-      # you can find examples here https://kubernetes.io/docs/concepts/services-networking/ingress-controllers
       # className: "nginx"
+      # and replace "nginx" by an Ingress controller name,
+      # examples [here](https://kubernetes.io/docs/concepts/services-networking/ingress-controllers).
       annotations: {}
         # kubernetes.io/ingress.class: nginx
         # kubernetes.io/tls-acme: "true"
       labels: {}
       hosts:
-       - host: "*.web.garage.tld" # wildcard website access with bucket name prefix
+      # --  wildcard website access with bucket name prefix
+       - host: "*.web.garage.tld"
          paths:
            - path: /
              pathType: Prefix
-       - host: "mywebpage.example.com" # specific bucket access with FQDN bucket
+       # -- specific bucket access with FQDN bucket
+       - host: "mywebpage.example.com"
          paths:
            - path: /
              pathType: Prefix
@@ -210,6 +197,21 @@ resources: {}
   #   cpu: 100m
   #   memory: 512Mi
 
+# -- Specifies a livenessProbe
+livenessProbe: {}
+  #httpGet:
+  #  path: /health
+  #  port: 3903
+  #initialDelaySeconds: 5
+  #periodSeconds: 30
+# -- Specifies a readinessProbe
+readinessProbe: {}
+  #httpGet:
+  #  path: /health
+  #  port: 3903
+  #initialDelaySeconds: 5
+  #periodSeconds: 30
+
 nodeSelector: {}
 
 tolerations: []
@@ -218,12 +220,16 @@ affinity: {}
 
 environment: {}
 
+extraVolumes: {}
+
+extraVolumeMounts: {}
+
 monitoring:
   metrics:
-    # If true, a service for monitoring is created with a prometheus.io/scrape annotation
+    # -- If true, a service for monitoring is created with a prometheus.io/scrape annotation
     enabled: false
     serviceMonitor:
-      # If true, a ServiceMonitor CRD is created for a prometheus operator
+      # -- If true, a ServiceMonitor CRD is created for a prometheus operator
       # https://github.com/coreos/prometheus-operator
       enabled: false
       path: /metrics
@@ -235,4 +241,5 @@ monitoring:
       scrapeTimeout: 10s
       relabelings: []
   tracing:
+    # -- specify a sink endpoint for OpenTelemetry Traces, eg. `http://localhost:4317`
     sink: ""

script/k8s/crd/garagenodes.deuxfleurs.fr.yaml

@@ -0,0 +1,43 @@
+apiVersion: apiextensions.k8s.io/v1
+kind: CustomResourceDefinition
+metadata:
+  name: garagenodes.deuxfleurs.fr
+spec:
+  conversion:
+    strategy: None
+  group: deuxfleurs.fr
+  names:
+    kind: GarageNode
+    listKind: GarageNodeList
+    plural: garagenodes
+    singular: garagenode
+  scope: Namespaced
+  versions:
+  - name: v1
+    schema:
+      openAPIV3Schema:
+        description: Auto-generated derived type for Node via `CustomResource`
+        properties:
+          spec:
+            properties:
+              address:
+                format: ip
+                type: string
+              hostname:
+                type: string
+              port:
+                format: uint16
+                minimum: 0
+                type: integer
+            required:
+            - address
+            - hostname
+            - port
+            type: object
+        required:
+        - spec
+        title: GarageNode
+        type: object
+    served: true
+    storage: true
+    subresources: {}

script/k8s/crd/kustomization.yaml

@@ -0,0 +1,5 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+
+resources:
+- garagenodes.deuxfleurs.fr.yaml

script/not-dynamic.sh

@@ -7,7 +7,12 @@ if [ "$#" -ne 1 ]; then
     exit 2
 fi
 
-if file $1 | grep 'dynamically linked' 2>&1; then
+if [ ! -x "$1" ]; then
+	echo "[fail] $1 does not exist or is not an executable"
+	exit 1
+fi
+
+if file "$1" | grep 'dynamically linked' 2>&1; then
     echo "[fail] $1 is dynamic"
     exit 1
 fi

shell.nix

@@ -3,7 +3,7 @@
 with import ./nix/common.nix;
 
 let
-  pkgs = import pkgsSrc {
+  pkgs = import nixpkgs {
     inherit system;
   };
   winscp = (import ./nix/winscp.nix) pkgs;
@@ -34,12 +34,14 @@ in
       jq
     ];
     shellHook = ''
+      export AWS_REQUEST_CHECKSUM_CALCULATION='when_required'
+
       function to_s3 {
         aws \
             --endpoint-url https://garage.deuxfleurs.fr \
             --region garage \
           s3 cp \
-            ./result-bin/bin/garage \
+            ./result/bin/garage \
             s3://garagehq.deuxfleurs.fr/_releases/''${CI_COMMIT_TAG:-$CI_COMMIT_SHA}/''${TARGET}/garage
       }
 
@@ -115,7 +117,7 @@ in
     shellHook = ''
       function refresh_cache {
         pass show deuxfleurs/nix_priv_key > /tmp/nix-signing-key.sec
-        for attr in clippy.amd64 test.amd64 pkgs.{amd64,i386,arm,arm64}.release; do
+        for attr in pkgs.amd64.debug test.amd64 pkgs.{amd64,i386,arm,arm64}.release; do
           echo "Updating cache for ''${attr}"
           nix copy -j8 \
             --to 's3://nix?endpoint=garage.deuxfleurs.fr&region=garage&secret-key=/tmp/nix-signing-key.sec' \

src/api/admin/Cargo.toml

@@ -0,0 +1,43 @@
+[package]
+name = "garage_api_admin"
+version = "1.3.1"
+authors = ["Alex Auvolat <alex@adnab.me>"]
+edition = "2018"
+license = "AGPL-3.0"
+description = "Admin API server crate for the Garage object store"
+repository = "https://git.deuxfleurs.fr/Deuxfleurs/garage"
+readme = "../../../README.md"
+
+[lib]
+path = "lib.rs"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+garage_model.workspace = true
+garage_table.workspace = true
+garage_util.workspace = true
+garage_rpc.workspace = true
+garage_api_common.workspace = true
+
+argon2.workspace = true
+async-trait.workspace = true
+thiserror.workspace = true
+hex.workspace = true
+tracing.workspace = true
+
+futures.workspace = true
+tokio.workspace = true
+http.workspace = true
+hyper = { workspace = true, default-features = false, features = ["server", "http1"] }
+url.workspace = true
+
+serde.workspace = true
+serde_json.workspace = true
+
+opentelemetry.workspace = true
+opentelemetry-prometheus = { workspace = true, optional = true }
+prometheus = { workspace = true, optional = true }
+
+[features]
+metrics = [ "opentelemetry-prometheus", "prometheus" ]

src/api/admin/api_server.rs

@@ -2,7 +2,6 @@ use std::collections::HashMap;
 use std::sync::Arc;
 
 use argon2::password_hash::PasswordHash;
-use async_trait::async_trait;
 
 use http::header::{ACCESS_CONTROL_ALLOW_METHODS, ACCESS_CONTROL_ALLOW_ORIGIN, ALLOW};
 use hyper::{body::Incoming as IncomingBody, Request, Response, StatusCode};
@@ -20,15 +19,15 @@ use garage_rpc::system::ClusterHealthStatus;
 use garage_util::error::Error as GarageError;
 use garage_util::socket_address::UnixOrTCPSocketAddress;
 
-use crate::generic_server::*;
+use garage_api_common::generic_server::*;
+use garage_api_common::helpers::*;
 
-use crate::admin::bucket::*;
+use crate::bucket::*;
-use crate::admin::cluster::*;
+use crate::cluster::*;
-use crate::admin::error::*;
+use crate::error::*;
-use crate::admin::key::*;
+use crate::key::*;
-use crate::admin::router_v0;
+use crate::router_v0;
-use crate::admin::router_v1::{Authorization, Endpoint};
+use crate::router_v1::{Authorization, Endpoint};
-use crate::helpers::*;
 
 pub type ResBody = BoxBody<Error>;
 
@@ -221,7 +220,6 @@ impl AdminApiServer {
 	}
 }
 
-#[async_trait]
 impl ApiHandler for AdminApiServer {
 	const API_NAME: &'static str = "admin";
 	const API_NAME_DISPLAY: &'static str = "Admin";

src/api/admin/bucket.rs

@@ -17,11 +17,12 @@ use garage_model::permission::*;
 use garage_model::s3::mpu_table;
 use garage_model::s3::object_table::*;
 
-use crate::admin::api_server::ResBody;
+use garage_api_common::common_error::CommonError;
-use crate::admin::error::*;
+use garage_api_common::helpers::*;
-use crate::admin::key::ApiBucketKeyPerm;
+
-use crate::common_error::CommonError;
+use crate::api_server::ResBody;
-use crate::helpers::*;
+use crate::error::*;
+use crate::key::ApiBucketKeyPerm;
 
 pub async fn handle_list_buckets(garage: &Arc<Garage>) -> Result<Response<ResBody>, Error> {
 	let buckets = garage
@@ -276,7 +277,7 @@ pub async fn handle_create_bucket(
 	let helper = garage.locked_helper().await;
 
 	if let Some(ga) = &req.global_alias {
-		if !is_valid_bucket_name(ga) {
+		if !is_valid_bucket_name(ga, garage.config.allow_punycode) {
 			return Err(Error::bad_request(format!(
 				"{}: {}",
 				ga, INVALID_BUCKET_NAME_MESSAGE
@@ -291,7 +292,7 @@ pub async fn handle_create_bucket(
 	}
 
 	if let Some(la) = &req.local_alias {
-		if !is_valid_bucket_name(&la.alias) {
+		if !is_valid_bucket_name(&la.alias, garage.config.allow_punycode) {
 			return Err(Error::bad_request(format!(
 				"{}: {}",
 				la.alias, INVALID_BUCKET_NAME_MESSAGE
@@ -381,7 +382,7 @@ pub async fn handle_delete_bucket(
 	for ((key_id, alias), _, active) in state.local_aliases.items().iter() {
 		if *active {
 			helper
-				.unset_local_bucket_alias(bucket.id, key_id, alias)
+				.purge_local_bucket_alias(bucket.id, key_id, alias)
 				.await?;
 		}
 	}

src/api/admin/cluster.rs

@@ -12,9 +12,10 @@ use garage_rpc::layout;
 
 use garage_model::garage::Garage;
 
-use crate::admin::api_server::ResBody;
+use garage_api_common::helpers::{json_ok_response, parse_json_body};
-use crate::admin::error::*;
+
-use crate::helpers::{json_ok_response, parse_json_body};
+use crate::api_server::ResBody;
+use crate::error::*;
 
 pub async fn handle_get_cluster_status(garage: &Arc<Garage>) -> Result<Response<ResBody>, Error> {
 	let layout = garage.system.cluster_layout();

src/api/admin/error.rs

@@ -1,45 +1,50 @@
-use err_derive::Error;
+use std::convert::TryFrom;
+
 use hyper::header::HeaderValue;
 use hyper::{HeaderMap, StatusCode};
+use thiserror::Error;
 
 pub use garage_model::helper::error::Error as HelperError;
 
-use crate::common_error::CommonError;
+use garage_api_common::common_error::{commonErrorDerivative, CommonError};
-pub use crate::common_error::{CommonErrorDerivative, OkOrBadRequest, OkOrInternalError};
+pub use garage_api_common::common_error::{
-use crate::generic_server::ApiError;
+	CommonErrorDerivative, OkOrBadRequest, OkOrInternalError,
-use crate::helpers::*;
+};
+use garage_api_common::generic_server::ApiError;
+use garage_api_common::helpers::*;
 
 /// Errors of this crate
 #[derive(Debug, Error)]
 pub enum Error {
-	#[error(display = "{}", _0)]
+	#[error("{0}")]
 	/// Error from common error
-	Common(CommonError),
+	Common(#[from] CommonError),
 
 	// Category: cannot process
 	/// The API access key does not exist
-	#[error(display = "Access key not found: {}", _0)]
+	#[error("Access key not found: {0}")]
 	NoSuchAccessKey(String),
 
 	/// In Import key, the key already exists
-	#[error(
+	#[error("Key {0} already exists in data store. Even if it is deleted, we can't let you create a new key with the same ID. Sorry.")]
-		display = "Key {} already exists in data store. Even if it is deleted, we can't let you create a new key with the same ID. Sorry.",
-		_0
-	)]
 	KeyAlreadyExists(String),
 }
 
-impl<T> From<T> for Error
+commonErrorDerivative!(Error);
-where
+
-	CommonError: From<T>,
+/// FIXME: helper errors are transformed into their corresponding variants
-{
+/// in the Error struct, but in many case a helper error should be considered
-	fn from(err: T) -> Self {
+/// an internal error.
-		Error::Common(CommonError::from(err))
+impl From<HelperError> for Error {
+	fn from(err: HelperError) -> Error {
+		match CommonError::try_from(err) {
+			Ok(ce) => Self::Common(ce),
+			Err(HelperError::NoSuchAccessKey(k)) => Self::NoSuchAccessKey(k),
+			Err(_) => unreachable!(),
+		}
 	}
 }
 
-impl CommonErrorDerivative for Error {}
-
 impl Error {
 	fn code(&self) -> &'static str {
 		match self {

src/api/admin/key.rs

@@ -9,9 +9,10 @@ use garage_table::*;
 use garage_model::garage::Garage;
 use garage_model::key_table::*;
 
-use crate::admin::api_server::ResBody;
+use garage_api_common::helpers::*;
-use crate::admin::error::*;
+
-use crate::helpers::*;
+use crate::api_server::ResBody;
+use crate::error::*;
 
 pub async fn handle_list_keys(garage: &Arc<Garage>) -> Result<Response<ResBody>, Error> {
 	let res = garage

src/api/admin/lib.rs

@@ -1,3 +1,6 @@
+#[macro_use]
+extern crate tracing;
+
 pub mod api_server;
 mod error;
 mod router_v0;

src/api/admin/router_v0.rs

@@ -2,8 +2,9 @@ use std::borrow::Cow;
 
 use hyper::{Method, Request};
 
-use crate::admin::error::*;
+use garage_api_common::router_macros::*;
-use crate::router_macros::*;
+
+use crate::error::*;
 
 router_match! {@func
 

src/api/admin/router_v1.rs

@@ -2,9 +2,10 @@ use std::borrow::Cow;
 
 use hyper::{Method, Request};
 
-use crate::admin::error::*;
+use garage_api_common::router_macros::*;
-use crate::admin::router_v0;
+
-use crate::router_macros::*;
+use crate::error::*;
+use crate::router_v0;
 
 pub enum Authorization {
 	None,

src/api/common/Cargo.toml

@@ -0,0 +1,48 @@
+[package]
+name = "garage_api_common"
+version = "1.3.1"
+authors = ["Alex Auvolat <alex@adnab.me>"]
+edition = "2018"
+license = "AGPL-3.0"
+description = "Common functions for the API server crates for the Garage object store"
+repository = "https://git.deuxfleurs.fr/Deuxfleurs/garage"
+readme = "../../../README.md"
+
+[lib]
+path = "lib.rs"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+garage_model.workspace = true
+garage_table.workspace = true
+garage_util.workspace = true
+
+base64.workspace = true
+bytes.workspace = true
+chrono.workspace = true
+crc32fast.workspace = true
+crc32c.workspace = true
+crypto-common.workspace = true
+thiserror.workspace = true
+hex.workspace = true
+hmac.workspace = true
+md-5.workspace = true
+tracing.workspace = true
+nom.workspace = true
+pin-project.workspace = true
+sha1.workspace = true
+sha2.workspace = true
+
+futures.workspace = true
+tokio.workspace = true
+http.workspace = true
+http-body-util.workspace = true
+hyper = { workspace = true, default-features = false, features = ["server", "http1"] }
+hyper-util.workspace = true
+url.workspace = true
+
+serde.workspace = true
+serde_json.workspace = true
+
+opentelemetry.workspace = true

src/api/common/common_error.rs

@@ -1,5 +1,7 @@
-use err_derive::Error;
+use std::convert::TryFrom;
+
 use hyper::StatusCode;
+use thiserror::Error;
 
 use garage_util::error::Error as GarageError;
 
@@ -10,51 +12,80 @@ use garage_model::helper::error::Error as HelperError;
 pub enum CommonError {
 	// ---- INTERNAL ERRORS ----
 	/// Error related to deeper parts of Garage
-	#[error(display = "Internal error: {}", _0)]
+	#[error("Internal error: {0}")]
-	InternalError(#[error(source)] GarageError),
+	InternalError(#[from] GarageError),
 
 	/// Error related to Hyper
-	#[error(display = "Internal error (Hyper error): {}", _0)]
+	#[error("Internal error (Hyper error): {0}")]
-	Hyper(#[error(source)] hyper::Error),
+	Hyper(#[from] hyper::Error),
 
 	/// Error related to HTTP
-	#[error(display = "Internal error (HTTP error): {}", _0)]
+	#[error("Internal error (HTTP error): {0}")]
-	Http(#[error(source)] http::Error),
+	Http(#[from] http::Error),
 
 	// ---- GENERIC CLIENT ERRORS ----
 	/// Proper authentication was not provided
-	#[error(display = "Forbidden: {}", _0)]
+	#[error("Forbidden: {0}")]
 	Forbidden(String),
 
 	/// Generic bad request response with custom message
-	#[error(display = "Bad request: {}", _0)]
+	#[error("Bad request: {0}")]
 	BadRequest(String),
 
 	/// The client sent a header with invalid value
-	#[error(display = "Invalid header value: {}", _0)]
+	#[error("Invalid header value: {0}")]
-	InvalidHeader(#[error(source)] hyper::header::ToStrError),
+	InvalidHeader(#[from] hyper::header::ToStrError),
 
 	// ---- SPECIFIC ERROR CONDITIONS ----
 	// These have to be error codes referenced in the S3 spec here:
 	// https://docs.aws.amazon.com/AmazonS3/latest/API/ErrorResponses.html#ErrorCodeList
 	/// The bucket requested don't exists
-	#[error(display = "Bucket not found: {}", _0)]
+	#[error("Bucket not found: {0}")]
 	NoSuchBucket(String),
 
 	/// Tried to create a bucket that already exist
-	#[error(display = "Bucket already exists")]
+	#[error("Bucket already exists")]
 	BucketAlreadyExists,
 
 	/// Tried to delete a non-empty bucket
-	#[error(display = "Tried to delete a non-empty bucket")]
+	#[error("Tried to delete a non-empty bucket")]
 	BucketNotEmpty,
 
 	// Category: bad request
 	/// Bucket name is not valid according to AWS S3 specs
-	#[error(display = "Invalid bucket name: {}", _0)]
+	#[error("Invalid bucket name: {0}")]
 	InvalidBucketName(String),
 }
 
+#[macro_export]
+macro_rules! commonErrorDerivative {
+	( $error_struct: ident ) => {
+		impl From<garage_util::error::Error> for $error_struct {
+			fn from(err: garage_util::error::Error) -> Self {
+				Self::Common(CommonError::InternalError(err))
+			}
+		}
+		impl From<http::Error> for $error_struct {
+			fn from(err: http::Error) -> Self {
+				Self::Common(CommonError::Http(err))
+			}
+		}
+		impl From<hyper::Error> for $error_struct {
+			fn from(err: hyper::Error) -> Self {
+				Self::Common(CommonError::Hyper(err))
+			}
+		}
+		impl From<hyper::header::ToStrError> for $error_struct {
+			fn from(err: hyper::header::ToStrError) -> Self {
+				Self::Common(CommonError::InvalidHeader(err))
+			}
+		}
+		impl CommonErrorDerivative for $error_struct {}
+	};
+}
+
+pub use commonErrorDerivative;
+
 impl CommonError {
 	pub fn http_status_code(&self) -> StatusCode {
 		match self {
@@ -97,18 +128,39 @@ impl CommonError {
 	}
 }
 
-impl From<HelperError> for CommonError {
+impl TryFrom<HelperError> for CommonError {
-	fn from(err: HelperError) -> Self {
+	type Error = HelperError;
+
+	fn try_from(err: HelperError) -> Result<Self, HelperError> {
 		match err {
-			HelperError::Internal(i) => Self::InternalError(i),
+			HelperError::Internal(i) => Ok(Self::InternalError(i)),
-			HelperError::BadRequest(b) => Self::BadRequest(b),
+			HelperError::BadRequest(b) => Ok(Self::BadRequest(b)),
-			HelperError::InvalidBucketName(n) => Self::InvalidBucketName(n),
+			HelperError::InvalidBucketName(n) => Ok(Self::InvalidBucketName(n)),
-			HelperError::NoSuchBucket(n) => Self::NoSuchBucket(n),
+			HelperError::NoSuchBucket(n) => Ok(Self::NoSuchBucket(n)),
-			e => Self::bad_request(format!("{}", e)),
+			e => Err(e),
 		}
 	}
 }
 
+/// This function converts HelperErrors into CommonErrors,
+/// for variants that exist in CommonError.
+/// This is used for helper functions that might return InvalidBucketName
+/// or NoSuchBucket for instance, and we want to pass that error
+/// up to our caller.
+pub fn pass_helper_error(err: HelperError) -> CommonError {
+	match CommonError::try_from(err) {
+		Ok(e) => e,
+		Err(e) => panic!("Helper error `{}` should hot have happenned here", e),
+	}
+}
+
+pub fn helper_error_as_internal(err: HelperError) -> CommonError {
+	match err {
+		HelperError::Internal(e) => CommonError::InternalError(e),
+		e => CommonError::InternalError(GarageError::Message(e.to_string())),
+	}
+}
+
 pub trait CommonErrorDerivative: From<CommonError> {
 	fn internal_error<M: ToString>(msg: M) -> Self {
 		Self::from(CommonError::InternalError(GarageError::Message(

src/api/common/cors.rs

@@ -0,0 +1,170 @@
+use std::sync::Arc;
+
+use http::header::{
+	ACCESS_CONTROL_ALLOW_HEADERS, ACCESS_CONTROL_ALLOW_METHODS, ACCESS_CONTROL_ALLOW_ORIGIN,
+	ACCESS_CONTROL_EXPOSE_HEADERS, ACCESS_CONTROL_REQUEST_HEADERS, ACCESS_CONTROL_REQUEST_METHOD,
+};
+use hyper::{body::Body, body::Incoming as IncomingBody, Request, Response, StatusCode};
+
+use garage_model::bucket_table::{BucketParams, CorsRule as GarageCorsRule};
+use garage_model::garage::Garage;
+
+use crate::common_error::{
+	helper_error_as_internal, CommonError, OkOrBadRequest, OkOrInternalError,
+};
+use crate::helpers::*;
+
+pub fn find_matching_cors_rule<'a, B>(
+	bucket_params: &'a BucketParams,
+	req: &Request<B>,
+) -> Result<Option<&'a GarageCorsRule>, CommonError> {
+	if let Some(cors_config) = bucket_params.cors_config.get() {
+		if let Some(origin) = req.headers().get("Origin") {
+			let origin = origin.to_str()?;
+			let request_headers = match req.headers().get(ACCESS_CONTROL_REQUEST_HEADERS) {
+				Some(h) => h.to_str()?.split(',').map(|h| h.trim()).collect::<Vec<_>>(),
+				None => vec![],
+			};
+			return Ok(cors_config.iter().find(|rule| {
+				cors_rule_matches(rule, origin, req.method().as_ref(), request_headers.iter())
+			}));
+		}
+	}
+	Ok(None)
+}
+
+pub fn cors_rule_matches<'a, HI, S>(
+	rule: &GarageCorsRule,
+	origin: &'a str,
+	method: &'a str,
+	mut request_headers: HI,
+) -> bool
+where
+	HI: Iterator<Item = S>,
+	S: AsRef<str>,
+{
+	rule.allow_origins.iter().any(|x| x == "*" || x == origin)
+		&& rule.allow_methods.iter().any(|x| x == "*" || x == method)
+		&& request_headers.all(|h| {
+			rule.allow_headers
+				.iter()
+				.any(|x| x == "*" || x == h.as_ref())
+		})
+}
+
+pub fn add_cors_headers(
+	resp: &mut Response<impl Body>,
+	rule: &GarageCorsRule,
+) -> Result<(), http::header::InvalidHeaderValue> {
+	let h = resp.headers_mut();
+	h.insert(
+		ACCESS_CONTROL_ALLOW_ORIGIN,
+		rule.allow_origins.join(", ").parse()?,
+	);
+	h.insert(
+		ACCESS_CONTROL_ALLOW_METHODS,
+		rule.allow_methods.join(", ").parse()?,
+	);
+	h.insert(
+		ACCESS_CONTROL_ALLOW_HEADERS,
+		rule.allow_headers.join(", ").parse()?,
+	);
+	h.insert(
+		ACCESS_CONTROL_EXPOSE_HEADERS,
+		rule.expose_headers.join(", ").parse()?,
+	);
+	Ok(())
+}
+
+pub async fn handle_options_api(
+	garage: Arc<Garage>,
+	req: &Request<IncomingBody>,
+	bucket_name: Option<String>,
+) -> Result<Response<EmptyBody>, CommonError> {
+	// FIXME: CORS rules of buckets with local aliases are
+	// not taken into account.
+
+	// If the bucket name is a global bucket name,
+	// we try to apply the CORS rules of that bucket.
+	// If a user has a local bucket name that has
+	// the same name, its CORS rules won't be applied
+	// and will be shadowed by the rules of the globally
+	// existing bucket (but this is inevitable because
+	// OPTIONS calls are not auhtenticated).
+	if let Some(bn) = bucket_name {
+		let helper = garage.bucket_helper();
+		let bucket_id = helper
+			.resolve_global_bucket_name(&bn)
+			.await
+			.map_err(helper_error_as_internal)?;
+		if let Some(id) = bucket_id {
+			let bucket = garage
+				.bucket_helper()
+				.get_existing_bucket(id)
+				.await
+				.map_err(helper_error_as_internal)?;
+			let bucket_params = bucket.state.into_option().unwrap();
+			handle_options_for_bucket(req, &bucket_params)
+		} else {
+			// If there is a bucket name in the request, but that name
+			// does not correspond to a global alias for a bucket,
+			// then it's either a non-existing bucket or a local bucket.
+			// We have no way of knowing, because the request is not
+			// authenticated and thus we can't resolve local aliases.
+			// We take the permissive approach of allowing everything,
+			// because we don't want to prevent web apps that use
+			// local bucket names from making API calls.
+			Ok(Response::builder()
+				.header(ACCESS_CONTROL_ALLOW_ORIGIN, "*")
+				.header(ACCESS_CONTROL_ALLOW_METHODS, "*")
+				.status(StatusCode::OK)
+				.body(EmptyBody::new())?)
+		}
+	} else {
+		// If there is no bucket name in the request,
+		// we are doing a ListBuckets call, which we want to allow
+		// for all origins.
+		Ok(Response::builder()
+			.header(ACCESS_CONTROL_ALLOW_ORIGIN, "*")
+			.header(ACCESS_CONTROL_ALLOW_METHODS, "GET")
+			.status(StatusCode::OK)
+			.body(EmptyBody::new())?)
+	}
+}
+
+pub fn handle_options_for_bucket<B>(
+	req: &Request<B>,
+	bucket_params: &BucketParams,
+) -> Result<Response<EmptyBody>, CommonError> {
+	let origin = req
+		.headers()
+		.get("Origin")
+		.ok_or_bad_request("Missing Origin header")?
+		.to_str()?;
+	let request_method = req
+		.headers()
+		.get(ACCESS_CONTROL_REQUEST_METHOD)
+		.ok_or_bad_request("Missing Access-Control-Request-Method header")?
+		.to_str()?;
+	let request_headers = match req.headers().get(ACCESS_CONTROL_REQUEST_HEADERS) {
+		Some(h) => h.to_str()?.split(',').map(|h| h.trim()).collect::<Vec<_>>(),
+		None => vec![],
+	};
+
+	if let Some(cors_config) = bucket_params.cors_config.get() {
+		let matching_rule = cors_config
+			.iter()
+			.find(|rule| cors_rule_matches(rule, origin, request_method, request_headers.iter()));
+		if let Some(rule) = matching_rule {
+			let mut resp = Response::builder()
+				.status(StatusCode::OK)
+				.body(EmptyBody::new())?;
+			add_cors_headers(&mut resp, rule).ok_or_internal_error("Invalid CORS configuration")?;
+			return Ok(resp);
+		}
+	}
+
+	Err(CommonError::Forbidden(
+		"This CORS request is not allowed.".into(),
+	))
+}

src/api/common/generic_server.rs

@@ -2,8 +2,7 @@ use std::convert::Infallible;
 use std::fs::{self, Permissions};
 use std::os::unix::fs::PermissionsExt;
 use std::sync::Arc;
-
+use std::time::Duration;
-use async_trait::async_trait;
 
 use futures::future::Future;
 use futures::stream::{futures_unordered::FuturesUnordered, StreamExt};
@@ -19,6 +18,7 @@ use hyper_util::rt::TokioIo;
 use tokio::io::{AsyncRead, AsyncWrite};
 use tokio::net::{TcpListener, TcpStream, UnixListener, UnixStream};
 use tokio::sync::watch;
+use tokio::time::{sleep_until, Instant};
 
 use opentelemetry::{
 	global,
@@ -34,7 +34,7 @@ use garage_util::socket_address::UnixOrTCPSocketAddress;
 
 use crate::helpers::{BoxBody, ErrorBody};
 
-pub(crate) trait ApiEndpoint: Send + Sync + 'static {
+pub trait ApiEndpoint: Send + Sync + 'static {
 	fn name(&self) -> &'static str;
 	fn add_span_attributes(&self, span: SpanRef<'_>);
 }
@@ -45,8 +45,7 @@ pub trait ApiError: std::error::Error + Send + Sync + 'static {
 	fn http_body(&self, garage_region: &str, path: &str) -> ErrorBody;
 }
 
-#[async_trait]
+pub trait ApiHandler: Send + Sync + 'static {
-pub(crate) trait ApiHandler: Send + Sync + 'static {
 	const API_NAME: &'static str;
 	const API_NAME_DISPLAY: &'static str;
 
@@ -54,14 +53,20 @@ pub(crate) trait ApiHandler: Send + Sync + 'static {
 	type Error: ApiError;
 
 	fn parse_endpoint(&self, r: &Request<IncomingBody>) -> Result<Self::Endpoint, Self::Error>;
-	async fn handle(
+	fn handle(
 		&self,
 		req: Request<IncomingBody>,
 		endpoint: Self::Endpoint,
-	) -> Result<Response<BoxBody<Self::Error>>, Self::Error>;
+	) -> impl Future<Output = Result<Response<BoxBody<Self::Error>>, Self::Error>> + Send;
+
+	/// Returns the key id used to authenticate this request. The ID returned must be safe to
+	/// log.
+	fn key_id_from_request(&self, _req: &Request<IncomingBody>) -> Option<String> {
+		None
+	}
 }
 
-pub(crate) struct ApiServer<A: ApiHandler> {
+pub struct ApiServer<A: ApiHandler> {
 	region: String,
 	api_handler: A,
 
@@ -143,19 +148,20 @@ impl<A: ApiHandler> ApiServer<A> {
 	) -> Result<Response<BoxBody<A::Error>>, http::Error> {
 		let uri = req.uri().clone();
 
-		if let Ok(forwarded_for_ip_addr) =
+		let source = if let Ok(forwarded_for_ip_addr) =
 			forwarded_headers::handle_forwarded_for_headers(req.headers())
 		{
-			info!(
+			format!("{forwarded_for_ip_addr} (via {addr})")
-				"{} (via {}) {} {}",
-				forwarded_for_ip_addr,
-				addr,
-				req.method(),
-				uri
-			);
 		} else {
-			info!("{} {} {}", addr, req.method(), uri);
+			format!("{addr}")
-		}
+		};
+		// we only do this to log the access key, so we can discard any error
+		let key = self
+			.api_handler
+			.key_id_from_request(&req)
+			.map(|k| format!("(key {k}) "))
+			.unwrap_or_default();
+		info!("{source} {key}{} {uri}", req.method());
 		debug!("{:?}", req);
 
 		let tracer = opentelemetry::global::tracer("garage");
@@ -246,13 +252,11 @@ impl<A: ApiHandler> ApiServer<A> {
 
 // ==== helper functions ====
 
-#[async_trait]
 pub trait Accept: Send + Sync + 'static {
 	type Stream: AsyncRead + AsyncWrite + Send + Sync + 'static;
-	async fn accept(&self) -> std::io::Result<(Self::Stream, String)>;
+	fn accept(&self) -> impl Future<Output = std::io::Result<(Self::Stream, String)>> + Send;
 }
 
-#[async_trait]
 impl Accept for TcpListener {
 	type Stream = TcpStream;
 	async fn accept(&self) -> std::io::Result<(Self::Stream, String)> {
@@ -264,7 +268,6 @@ impl Accept for TcpListener {
 
 pub struct UnixListenerOn(pub UnixListener, pub String);
 
-#[async_trait]
 impl Accept for UnixListenerOn {
 	type Stream = UnixStream;
 	async fn accept(&self) -> std::io::Result<(Self::Stream, String)> {
@@ -291,7 +294,7 @@ where
 	let connection_collector = tokio::spawn({
 		let server_name = server_name.clone();
 		async move {
-			let mut connections = FuturesUnordered::new();
+			let mut connections = FuturesUnordered::<tokio::task::JoinHandle<()>>::new();
 			loop {
 				let collect_next = async {
 					if connections.is_empty() {
@@ -312,23 +315,34 @@ where
 					}
 				}
 			}
-			if !connections.is_empty() {
+			let deadline = Instant::now() + Duration::from_secs(10);
+			while !connections.is_empty() {
 				info!(
-					"{} server: {} connections still open",
+					"{} server: {} connections still open, deadline in {:.2}s",
 					server_name,
-					connections.len()
+					connections.len(),
+					(deadline - Instant::now()).as_secs_f32(),
 				);
-				while let Some(conn_res) = connections.next().await {
+				tokio::select! {
-					trace!(
+					conn_res = connections.next() => {
-						"{} server: HTTP connection finished: {:?}",
+						trace!(
-						server_name,
+							"{} server: HTTP connection finished: {:?}",
-						conn_res
+							server_name,
-					);
+							conn_res.unwrap(),
-					info!(
+						);
-						"{} server: {} connections still open",
+					}
-						server_name,
+					_ = sleep_until(deadline) => {
-						connections.len()
+						warn!("{} server: exit deadline reached with {} connections still open, killing them now",
-					);
+							server_name,
+							connections.len());
+						for conn in connections.iter() {
+							conn.abort();
+						}
+						for conn in connections {
+							assert!(conn.await.unwrap_err().is_cancelled());
+						}
+						break;
+					}
 				}
 			}
 		}
@@ -336,7 +350,11 @@ where
 
 	while !*must_exit.borrow() {
 		let (stream, client_addr) = tokio::select! {
-			acc = listener.accept() => acc?,
+			acc = listener.accept() => match acc {
+							Ok(r) => r,
+							Err(e) if e.kind() == std::io::ErrorKind::ConnectionAborted => continue,
+							Err(e) => return Err(e.into()),
+						},
 			_ = must_exit.changed() => continue,
 		};
 

src/api/common/helpers.rs

@@ -8,7 +8,6 @@ use hyper::{
 	body::{Body, Bytes},
 	Request, Response,
 };
-use idna::domain_to_unicode;
 use serde::{Deserialize, Serialize};
 
 use garage_model::bucket_table::BucketParams;
@@ -97,7 +96,7 @@ pub fn authority_to_host(authority: &str) -> Result<String, Error> {
 			authority
 		))),
 	};
-	authority.map(|h| domain_to_unicode(h).0)
+	authority.map(|h| h.to_ascii_lowercase())
 }
 
 /// Extract the bucket name and the key name from an HTTP path and possibly a bucket provided in
@@ -363,9 +362,9 @@ mod tests {
 }
 
 #[derive(Serialize)]
-pub(crate) struct CustomApiErrorBody {
+pub struct CustomApiErrorBody {
-	pub(crate) code: String,
+	pub code: String,
-	pub(crate) message: String,
+	pub message: String,
-	pub(crate) region: String,
+	pub region: String,
-	pub(crate) path: String,
+	pub path: String,
 }

src/api/common/lib.rs

@@ -0,0 +1,12 @@
+//! Crate for serving a S3 compatible API
+#[macro_use]
+extern crate tracing;
+
+pub mod common_error;
+
+pub mod cors;
+pub mod encoding;
+pub mod generic_server;
+pub mod helpers;
+pub mod router_macros;
+pub mod signature;

src/api/common/router_macros.rs

@@ -1,5 +1,6 @@
 /// This macro is used to generate very repetitive match {} blocks in this module
 /// It is _not_ made to be used anywhere else
+#[macro_export]
 macro_rules! router_match {
     (@match $enum:expr , [ $($endpoint:ident,)* ]) => {{
         // usage: router_match {@match my_enum, [ VariantWithField1, VariantWithField2 ..] }
@@ -133,6 +134,7 @@ macro_rules! router_match {
 
 /// This macro is used to generate part of the code in this module. It must be called only one, and
 /// is useless outside of this module.
+#[macro_export]
 macro_rules! generateQueryParameters {
     (
         keywords: [ $($kw_param:expr => $kw_name: ident),* ],
@@ -204,7 +206,7 @@ macro_rules! generateQueryParameters {
             }
 
             /// Get an error message in case not all parameters where used when extracting them to
-            /// build an Enpoint variant
+            /// build an Endpoint variant
             fn nonempty_message(&self) -> Option<&str> {
                 if self.keyword.is_some() {
                     Some("Keyword not used")
@@ -220,5 +222,5 @@ macro_rules! generateQueryParameters {
     }
 }
 
-pub(crate) use generateQueryParameters;
+pub use generateQueryParameters;
-pub(crate) use router_match;
+pub use router_match;

src/api/common/signature/body.rs

@@ -0,0 +1,135 @@
+use std::sync::Mutex;
+
+use futures::prelude::*;
+use futures::stream::BoxStream;
+use http_body_util::{BodyExt, StreamBody};
+use hyper::body::{Bytes, Frame};
+use serde::Deserialize;
+use tokio::sync::mpsc;
+use tokio::task;
+
+use super::*;
+
+use crate::signature::checksum::*;
+
+pub struct ReqBody {
+	// why need mutex to be sync??
+	pub(crate) stream: Mutex<BoxStream<'static, Result<Frame<Bytes>, Error>>>,
+	pub(crate) checksummer: Checksummer,
+	pub(crate) expected_checksums: ExpectedChecksums,
+	pub(crate) trailer_algorithm: Option<ChecksumAlgorithm>,
+}
+
+pub type StreamingChecksumReceiver = task::JoinHandle<Result<Checksums, Error>>;
+
+impl ReqBody {
+	pub fn add_expected_checksums(&mut self, more: ExpectedChecksums) {
+		if more.md5.is_some() {
+			self.expected_checksums.md5 = more.md5;
+		}
+		if more.sha256.is_some() {
+			self.expected_checksums.sha256 = more.sha256;
+		}
+		if more.extra.is_some() {
+			self.expected_checksums.extra = more.extra;
+		}
+		self.checksummer.add_expected(&self.expected_checksums);
+	}
+
+	pub fn add_md5(&mut self) {
+		self.checksummer.add_md5();
+	}
+
+	// ============ non-streaming =============
+
+	pub async fn json<T: for<'a> Deserialize<'a>>(self) -> Result<T, Error> {
+		let body = self.collect().await?;
+		let resp: T = serde_json::from_slice(&body).ok_or_bad_request("Invalid JSON")?;
+		Ok(resp)
+	}
+
+	pub async fn collect(self) -> Result<Bytes, Error> {
+		self.collect_with_checksums().await.map(|(b, _)| b)
+	}
+
+	pub async fn collect_with_checksums(mut self) -> Result<(Bytes, Checksums), Error> {
+		let stream: BoxStream<_> = self.stream.into_inner().unwrap();
+		let bytes = BodyExt::collect(StreamBody::new(stream)).await?.to_bytes();
+
+		self.checksummer.update(&bytes);
+		let checksums = self.checksummer.finalize();
+		checksums.verify(&self.expected_checksums)?;
+
+		Ok((bytes, checksums))
+	}
+
+	// ============ streaming =============
+
+	pub fn streaming_with_checksums(
+		self,
+	) -> (
+		BoxStream<'static, Result<Bytes, Error>>,
+		StreamingChecksumReceiver,
+	) {
+		let Self {
+			stream,
+			mut checksummer,
+			mut expected_checksums,
+			trailer_algorithm,
+		} = self;
+
+		let (frame_tx, mut frame_rx) = mpsc::channel::<Frame<Bytes>>(5);
+
+		let join_checksums = tokio::spawn(async move {
+			while let Some(frame) = frame_rx.recv().await {
+				match frame.into_data() {
+					Ok(data) => {
+						checksummer = tokio::task::spawn_blocking(move || {
+							checksummer.update(&data);
+							checksummer
+						})
+						.await
+						.unwrap()
+					}
+					Err(frame) => {
+						let trailers = frame.into_trailers().unwrap();
+						let algo = trailer_algorithm.unwrap();
+						expected_checksums.extra = Some(extract_checksum_value(&trailers, algo)?);
+						break;
+					}
+				}
+			}
+
+			if trailer_algorithm.is_some() && expected_checksums.extra.is_none() {
+				return Err(Error::bad_request("trailing checksum was not sent"));
+			}
+
+			let checksums = checksummer.finalize();
+			checksums.verify(&expected_checksums)?;
+
+			Ok(checksums)
+		});
+
+		let stream: BoxStream<_> = stream.into_inner().unwrap();
+		let stream = stream.filter_map(move |x| {
+			let frame_tx = frame_tx.clone();
+			async move {
+				match x {
+					Err(e) => Some(Err(e)),
+					Ok(frame) => {
+						if frame.is_data() {
+							let data = frame.data_ref().unwrap().clone();
+							let _ = frame_tx.send(frame).await;
+							Some(Ok(data))
+						} else {
+							let _ = frame_tx.send(frame).await;
+							None
+						}
+					}
+				}
+			}
+		});
+
+		(stream.boxed(), join_checksums)
+	}
+}

src/api/common/signature/checksum.rs

@@ -11,11 +11,12 @@ use sha2::Sha256;
 use http::{HeaderMap, HeaderName, HeaderValue};
 
 use garage_util::data::*;
-use garage_util::error::OkOrMessage;
 
-use garage_model::s3::object_table::*;
+use super::*;
 
-use crate::s3::error::*;
+pub use garage_model::s3::object_table::{ChecksumAlgorithm, ChecksumValue};
+
+pub const CONTENT_MD5: HeaderName = HeaderName::from_static("content-md5");
 
 pub const X_AMZ_CHECKSUM_ALGORITHM: HeaderName =
 	HeaderName::from_static("x-amz-checksum-algorithm");
@@ -31,8 +32,8 @@ pub type Md5Checksum = [u8; 16];
 pub type Sha1Checksum = [u8; 20];
 pub type Sha256Checksum = [u8; 32];
 
-#[derive(Debug, Default)]
+#[derive(Debug, Default, Clone)]
-pub(crate) struct ExpectedChecksums {
+pub struct ExpectedChecksums {
 	// base64-encoded md5 (content-md5 header)
 	pub md5: Option<String>,
 	// content_sha256 (as a Hash / FixedBytes32)
@@ -41,7 +42,7 @@ pub(crate) struct ExpectedChecksums {
 	pub extra: Option<ChecksumValue>,
 }
 
-pub(crate) struct Checksummer {
+pub struct Checksummer {
 	pub crc32: Option<Crc32>,
 	pub crc32c: Option<Crc32c>,
 	pub md5: Option<Md5>,
@@ -50,7 +51,7 @@ pub(crate) struct Checksummer {
 }
 
 #[derive(Default)]
-pub(crate) struct Checksums {
+pub struct Checksums {
 	pub crc32: Option<Crc32Checksum>,
 	pub crc32c: Option<Crc32cChecksum>,
 	pub md5: Option<Md5Checksum>,
@@ -59,34 +60,48 @@ pub(crate) struct Checksums {
 }
 
 impl Checksummer {
-	pub(crate) fn init(expected: &ExpectedChecksums, require_md5: bool) -> Self {
+	pub fn new() -> Self {
-		let mut ret = Self {
+		Self {
 			crc32: None,
 			crc32c: None,
 			md5: None,
 			sha1: None,
 			sha256: None,
-		};
+		}
+	}
 
-		if expected.md5.is_some() || require_md5 {
+	pub fn init(expected: &ExpectedChecksums, add_md5: bool) -> Self {
-			ret.md5 = Some(Md5::new());
+		let mut ret = Self::new();
-		}
+		ret.add_expected(expected);
-		if expected.sha256.is_some() || matches!(&expected.extra, Some(ChecksumValue::Sha256(_))) {
+		if add_md5 {
-			ret.sha256 = Some(Sha256::new());
+			ret.add_md5();
-		}
-		if matches!(&expected.extra, Some(ChecksumValue::Crc32(_))) {
-			ret.crc32 = Some(Crc32::new());
-		}
-		if matches!(&expected.extra, Some(ChecksumValue::Crc32c(_))) {
-			ret.crc32c = Some(Crc32c::default());
-		}
-		if matches!(&expected.extra, Some(ChecksumValue::Sha1(_))) {
-			ret.sha1 = Some(Sha1::new());
 		}
 		ret
 	}
 
-	pub(crate) fn add(mut self, algo: Option<ChecksumAlgorithm>) -> Self {
+	pub fn add_md5(&mut self) {
+		self.md5 = Some(Md5::new());
+	}
+
+	pub fn add_expected(&mut self, expected: &ExpectedChecksums) {
+		if expected.md5.is_some() {
+			self.md5 = Some(Md5::new());
+		}
+		if expected.sha256.is_some() || matches!(&expected.extra, Some(ChecksumValue::Sha256(_))) {
+			self.sha256 = Some(Sha256::new());
+		}
+		if matches!(&expected.extra, Some(ChecksumValue::Crc32(_))) {
+			self.crc32 = Some(Crc32::new());
+		}
+		if matches!(&expected.extra, Some(ChecksumValue::Crc32c(_))) {
+			self.crc32c = Some(Crc32c::default());
+		}
+		if matches!(&expected.extra, Some(ChecksumValue::Sha1(_))) {
+			self.sha1 = Some(Sha1::new());
+		}
+	}
+
+	pub fn add(mut self, algo: Option<ChecksumAlgorithm>) -> Self {
 		match algo {
 			Some(ChecksumAlgorithm::Crc32) => {
 				self.crc32 = Some(Crc32::new());
@@ -105,7 +120,7 @@ impl Checksummer {
 		self
 	}
 
-	pub(crate) fn update(&mut self, bytes: &[u8]) {
+	pub fn update(&mut self, bytes: &[u8]) {
 		if let Some(crc32) = &mut self.crc32 {
 			crc32.update(bytes);
 		}
@@ -123,7 +138,7 @@ impl Checksummer {
 		}
 	}
 
-	pub(crate) fn finalize(self) -> Checksums {
+	pub fn finalize(self) -> Checksums {
 		Checksums {
 			crc32: self.crc32.map(|x| u32::to_be_bytes(x.finalize())),
 			crc32c: self
@@ -183,153 +198,56 @@ impl Checksums {
 
 // ----
 
-#[derive(Default)]
+pub fn parse_checksum_algorithm(algo: &str) -> Result<ChecksumAlgorithm, Error> {
-pub(crate) struct MultipartChecksummer {
+	match algo {
-	pub md5: Md5,
+		"CRC32" => Ok(ChecksumAlgorithm::Crc32),
-	pub extra: Option<MultipartExtraChecksummer>,
+		"CRC32C" => Ok(ChecksumAlgorithm::Crc32c),
-}
+		"SHA1" => Ok(ChecksumAlgorithm::Sha1),
-
+		"SHA256" => Ok(ChecksumAlgorithm::Sha256),
-pub(crate) enum MultipartExtraChecksummer {
+		_ => Err(Error::bad_request("invalid checksum algorithm")),
-	Crc32(Crc32),
-	Crc32c(Crc32c),
-	Sha1(Sha1),
-	Sha256(Sha256),
-}
-
-impl MultipartChecksummer {
-	pub(crate) fn init(algo: Option<ChecksumAlgorithm>) -> Self {
-		Self {
-			md5: Md5::new(),
-			extra: match algo {
-				None => None,
-				Some(ChecksumAlgorithm::Crc32) => {
-					Some(MultipartExtraChecksummer::Crc32(Crc32::new()))
-				}
-				Some(ChecksumAlgorithm::Crc32c) => {
-					Some(MultipartExtraChecksummer::Crc32c(Crc32c::default()))
-				}
-				Some(ChecksumAlgorithm::Sha1) => Some(MultipartExtraChecksummer::Sha1(Sha1::new())),
-				Some(ChecksumAlgorithm::Sha256) => {
-					Some(MultipartExtraChecksummer::Sha256(Sha256::new()))
-				}
-			},
-		}
-	}
-
-	pub(crate) fn update(
-		&mut self,
-		etag: &str,
-		checksum: Option<ChecksumValue>,
-	) -> Result<(), Error> {
-		self.md5
-			.update(&hex::decode(&etag).ok_or_message("invalid etag hex")?);
-		match (&mut self.extra, checksum) {
-			(None, _) => (),
-			(
-				Some(MultipartExtraChecksummer::Crc32(ref mut crc32)),
-				Some(ChecksumValue::Crc32(x)),
-			) => {
-				crc32.update(&x);
-			}
-			(
-				Some(MultipartExtraChecksummer::Crc32c(ref mut crc32c)),
-				Some(ChecksumValue::Crc32c(x)),
-			) => {
-				crc32c.write(&x);
-			}
-			(Some(MultipartExtraChecksummer::Sha1(ref mut sha1)), Some(ChecksumValue::Sha1(x))) => {
-				sha1.update(&x);
-			}
-			(
-				Some(MultipartExtraChecksummer::Sha256(ref mut sha256)),
-				Some(ChecksumValue::Sha256(x)),
-			) => {
-				sha256.update(&x);
-			}
-			(Some(_), b) => {
-				return Err(Error::internal_error(format!(
-					"part checksum was not computed correctly, got: {:?}",
-					b
-				)))
-			}
-		}
-		Ok(())
-	}
-
-	pub(crate) fn finalize(self) -> (Md5Checksum, Option<ChecksumValue>) {
-		let md5 = self.md5.finalize()[..].try_into().unwrap();
-		let extra = match self.extra {
-			None => None,
-			Some(MultipartExtraChecksummer::Crc32(crc32)) => {
-				Some(ChecksumValue::Crc32(u32::to_be_bytes(crc32.finalize())))
-			}
-			Some(MultipartExtraChecksummer::Crc32c(crc32c)) => Some(ChecksumValue::Crc32c(
-				u32::to_be_bytes(u32::try_from(crc32c.finish()).unwrap()),
-			)),
-			Some(MultipartExtraChecksummer::Sha1(sha1)) => {
-				Some(ChecksumValue::Sha1(sha1.finalize()[..].try_into().unwrap()))
-			}
-			Some(MultipartExtraChecksummer::Sha256(sha256)) => Some(ChecksumValue::Sha256(
-				sha256.finalize()[..].try_into().unwrap(),
-			)),
-		};
-		(md5, extra)
 	}
 }
 
-// ----
-
 /// Extract the value of the x-amz-checksum-algorithm header
-pub(crate) fn request_checksum_algorithm(
+pub fn request_checksum_algorithm(
 	headers: &HeaderMap<HeaderValue>,
 ) -> Result<Option<ChecksumAlgorithm>, Error> {
 	match headers.get(X_AMZ_CHECKSUM_ALGORITHM) {
 		None => Ok(None),
-		Some(x) if x == "CRC32" => Ok(Some(ChecksumAlgorithm::Crc32)),
+		Some(x) => parse_checksum_algorithm(x.to_str()?).map(Some),
-		Some(x) if x == "CRC32C" => Ok(Some(ChecksumAlgorithm::Crc32c)),
+	}
-		Some(x) if x == "SHA1" => Ok(Some(ChecksumAlgorithm::Sha1)),
+}
-		Some(x) if x == "SHA256" => Ok(Some(ChecksumAlgorithm::Sha256)),
+
+pub fn request_trailer_checksum_algorithm(
+	headers: &HeaderMap<HeaderValue>,
+) -> Result<Option<ChecksumAlgorithm>, Error> {
+	match headers.get(X_AMZ_TRAILER).map(|x| x.to_str()).transpose()? {
+		None => Ok(None),
+		Some(x) if x == X_AMZ_CHECKSUM_CRC32 => Ok(Some(ChecksumAlgorithm::Crc32)),
+		Some(x) if x == X_AMZ_CHECKSUM_CRC32C => Ok(Some(ChecksumAlgorithm::Crc32c)),
+		Some(x) if x == X_AMZ_CHECKSUM_SHA1 => Ok(Some(ChecksumAlgorithm::Sha1)),
+		Some(x) if x == X_AMZ_CHECKSUM_SHA256 => Ok(Some(ChecksumAlgorithm::Sha256)),
 		_ => Err(Error::bad_request("invalid checksum algorithm")),
 	}
 }
 
 /// Extract the value of any of the x-amz-checksum-* headers
-pub(crate) fn request_checksum_value(
+pub fn request_checksum_value(
 	headers: &HeaderMap<HeaderValue>,
 ) -> Result<Option<ChecksumValue>, Error> {
 	let mut ret = vec![];
 
-	if let Some(crc32_str) = headers.get(X_AMZ_CHECKSUM_CRC32) {
+	if headers.contains_key(X_AMZ_CHECKSUM_CRC32) {
-		let crc32 = BASE64_STANDARD
+		ret.push(extract_checksum_value(headers, ChecksumAlgorithm::Crc32)?);
-			.decode(&crc32_str)
-			.ok()
-			.and_then(|x| x.try_into().ok())
-			.ok_or_bad_request("invalid x-amz-checksum-crc32 header")?;
-		ret.push(ChecksumValue::Crc32(crc32))
 	}
-	if let Some(crc32c_str) = headers.get(X_AMZ_CHECKSUM_CRC32C) {
+	if headers.contains_key(X_AMZ_CHECKSUM_CRC32C) {
-		let crc32c = BASE64_STANDARD
+		ret.push(extract_checksum_value(headers, ChecksumAlgorithm::Crc32c)?);
-			.decode(&crc32c_str)
-			.ok()
-			.and_then(|x| x.try_into().ok())
-			.ok_or_bad_request("invalid x-amz-checksum-crc32c header")?;
-		ret.push(ChecksumValue::Crc32c(crc32c))
 	}
-	if let Some(sha1_str) = headers.get(X_AMZ_CHECKSUM_SHA1) {
+	if headers.contains_key(X_AMZ_CHECKSUM_SHA1) {
-		let sha1 = BASE64_STANDARD
+		ret.push(extract_checksum_value(headers, ChecksumAlgorithm::Sha1)?);
-			.decode(&sha1_str)
-			.ok()
-			.and_then(|x| x.try_into().ok())
-			.ok_or_bad_request("invalid x-amz-checksum-sha1 header")?;
-		ret.push(ChecksumValue::Sha1(sha1))
 	}
-	if let Some(sha256_str) = headers.get(X_AMZ_CHECKSUM_SHA256) {
+	if headers.contains_key(X_AMZ_CHECKSUM_SHA256) {
-		let sha256 = BASE64_STANDARD
+		ret.push(extract_checksum_value(headers, ChecksumAlgorithm::Sha256)?);
-			.decode(&sha256_str)
-			.ok()
-			.and_then(|x| x.try_into().ok())
-			.ok_or_bad_request("invalid x-amz-checksum-sha256 header")?;
-		ret.push(ChecksumValue::Sha256(sha256))
 	}
 
 	if ret.len() > 1 {
@@ -340,50 +258,49 @@ pub(crate) fn request_checksum_value(
 	Ok(ret.pop())
 }
 
-/// Checks for the presense of x-amz-checksum-algorithm
+/// Checks for the presence of x-amz-checksum-algorithm
-/// if so extract the corrseponding x-amz-checksum-* value
+/// if so extract the corresponding x-amz-checksum-* value
-pub(crate) fn request_checksum_algorithm_value(
+pub fn extract_checksum_value(
 	headers: &HeaderMap<HeaderValue>,
-) -> Result<Option<ChecksumValue>, Error> {
+	algo: ChecksumAlgorithm,
-	match headers.get(X_AMZ_CHECKSUM_ALGORITHM) {
+) -> Result<ChecksumValue, Error> {
-		Some(x) if x == "CRC32" => {
+	match algo {
+		ChecksumAlgorithm::Crc32 => {
 			let crc32 = headers
 				.get(X_AMZ_CHECKSUM_CRC32)
 				.and_then(|x| BASE64_STANDARD.decode(&x).ok())
 				.and_then(|x| x.try_into().ok())
 				.ok_or_bad_request("invalid x-amz-checksum-crc32 header")?;
-			Ok(Some(ChecksumValue::Crc32(crc32)))
+			Ok(ChecksumValue::Crc32(crc32))
 		}
-		Some(x) if x == "CRC32C" => {
+		ChecksumAlgorithm::Crc32c => {
 			let crc32c = headers
 				.get(X_AMZ_CHECKSUM_CRC32C)
 				.and_then(|x| BASE64_STANDARD.decode(&x).ok())
 				.and_then(|x| x.try_into().ok())
 				.ok_or_bad_request("invalid x-amz-checksum-crc32c header")?;
-			Ok(Some(ChecksumValue::Crc32c(crc32c)))
+			Ok(ChecksumValue::Crc32c(crc32c))
 		}
-		Some(x) if x == "SHA1" => {
+		ChecksumAlgorithm::Sha1 => {
 			let sha1 = headers
 				.get(X_AMZ_CHECKSUM_SHA1)
 				.and_then(|x| BASE64_STANDARD.decode(&x).ok())
 				.and_then(|x| x.try_into().ok())
 				.ok_or_bad_request("invalid x-amz-checksum-sha1 header")?;
-			Ok(Some(ChecksumValue::Sha1(sha1)))
+			Ok(ChecksumValue::Sha1(sha1))
 		}
-		Some(x) if x == "SHA256" => {
+		ChecksumAlgorithm::Sha256 => {
 			let sha256 = headers
 				.get(X_AMZ_CHECKSUM_SHA256)
 				.and_then(|x| BASE64_STANDARD.decode(&x).ok())
 				.and_then(|x| x.try_into().ok())
 				.ok_or_bad_request("invalid x-amz-checksum-sha256 header")?;
-			Ok(Some(ChecksumValue::Sha256(sha256)))
+			Ok(ChecksumValue::Sha256(sha256))
 		}
-		Some(_) => Err(Error::bad_request("invalid x-amz-checksum-algorithm")),
-		None => Ok(None),
 	}
 }
 
-pub(crate) fn add_checksum_response_headers(
+pub fn add_checksum_response_headers(
 	checksum: &Option<ChecksumValue>,
 	mut resp: http::response::Builder,
 ) -> http::response::Builder {

src/api/common/signature/error.rs

@@ -1,4 +1,4 @@
-use err_derive::Error;
+use thiserror::Error;
 
 use crate::common_error::CommonError;
 pub use crate::common_error::{CommonErrorDerivative, OkOrBadRequest, OkOrInternalError};
@@ -6,18 +6,22 @@ pub use crate::common_error::{CommonErrorDerivative, OkOrBadRequest, OkOrInterna
 /// Errors of this crate
 #[derive(Debug, Error)]
 pub enum Error {
-	#[error(display = "{}", _0)]
+	#[error("{0}")]
 	/// Error from common error
 	Common(CommonError),
 
 	/// Authorization Header Malformed
-	#[error(display = "Authorization header malformed, unexpected scope: {}", _0)]
+	#[error("Authorization header malformed, unexpected scope: {0}")]
 	AuthorizationHeaderMalformed(String),
 
 	// Category: bad request
 	/// The request contained an invalid UTF-8 sequence in its path or in other parameters
-	#[error(display = "Invalid UTF-8: {}", _0)]
+	#[error("Invalid UTF-8: {0}")]
-	InvalidUtf8Str(#[error(source)] std::str::Utf8Error),
+	InvalidUtf8Str(#[from] std::str::Utf8Error),
+
+	/// The provided digest (checksum) value was invalid
+	#[error("Invalid digest: {0}")]
+	InvalidDigest(String),
 }
 
 impl<T> From<T> for Error

src/api/common/signature/mod.rs

@@ -0,0 +1,118 @@
+use chrono::{DateTime, Utc};
+use hmac::{Hmac, Mac};
+use sha2::Sha256;
+
+use hyper::header::HeaderName;
+use hyper::{body::Incoming as IncomingBody, Request};
+
+use garage_model::garage::Garage;
+use garage_model::key_table::Key;
+use garage_util::data::{sha256sum, Hash};
+
+use error::*;
+
+pub mod body;
+pub mod checksum;
+pub mod error;
+pub mod payload;
+pub mod streaming;
+
+pub const SHORT_DATE: &str = "%Y%m%d";
+pub const LONG_DATETIME: &str = "%Y%m%dT%H%M%SZ";
+
+// ---- Constants used in AWSv4 signatures ----
+
+pub const X_AMZ_ALGORITHM: HeaderName = HeaderName::from_static("x-amz-algorithm");
+pub const X_AMZ_CREDENTIAL: HeaderName = HeaderName::from_static("x-amz-credential");
+pub const X_AMZ_DATE: HeaderName = HeaderName::from_static("x-amz-date");
+pub const X_AMZ_EXPIRES: HeaderName = HeaderName::from_static("x-amz-expires");
+pub const X_AMZ_SIGNEDHEADERS: HeaderName = HeaderName::from_static("x-amz-signedheaders");
+pub const X_AMZ_SIGNATURE: HeaderName = HeaderName::from_static("x-amz-signature");
+pub const X_AMZ_CONTENT_SHA256: HeaderName = HeaderName::from_static("x-amz-content-sha256");
+pub const X_AMZ_TRAILER: HeaderName = HeaderName::from_static("x-amz-trailer");
+
+/// Result of `sha256("")`
+pub(crate) const EMPTY_STRING_HEX_DIGEST: &str =
+	"e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855";
+
+// Signature calculation algorithm
+pub const AWS4_HMAC_SHA256: &str = "AWS4-HMAC-SHA256";
+type HmacSha256 = Hmac<Sha256>;
+
+// Possible values for x-amz-content-sha256, in addition to the actual sha256
+pub const UNSIGNED_PAYLOAD: &str = "UNSIGNED-PAYLOAD";
+pub const STREAMING_UNSIGNED_PAYLOAD_TRAILER: &str = "STREAMING-UNSIGNED-PAYLOAD-TRAILER";
+pub const STREAMING_AWS4_HMAC_SHA256_PAYLOAD: &str = "STREAMING-AWS4-HMAC-SHA256-PAYLOAD";
+
+// Used in the computation of StringToSign
+pub const AWS4_HMAC_SHA256_PAYLOAD: &str = "AWS4-HMAC-SHA256-PAYLOAD";
+
+// ---- enums to describe stuff going on in signature calculation ----
+
+#[derive(Debug)]
+pub enum ContentSha256Header {
+	UnsignedPayload,
+	Sha256Checksum(Hash),
+	StreamingPayload { trailer: bool, signed: bool },
+}
+
+// ---- top-level functions ----
+
+pub struct VerifiedRequest {
+	pub request: Request<streaming::ReqBody>,
+	pub access_key: Key,
+	pub content_sha256_header: ContentSha256Header,
+}
+
+pub async fn verify_request(
+	garage: &Garage,
+	mut req: Request<IncomingBody>,
+	service: &'static str,
+) -> Result<VerifiedRequest, Error> {
+	let checked_signature = payload::check_payload_signature(&garage, &mut req, service).await?;
+
+	let request = streaming::parse_streaming_body(
+		req,
+		&checked_signature,
+		&garage.config.s3_api.s3_region,
+		service,
+	)?;
+
+	let access_key = checked_signature
+		.key
+		.ok_or_else(|| Error::forbidden("Garage does not support anonymous access yet"))?;
+
+	Ok(VerifiedRequest {
+		request,
+		access_key,
+		content_sha256_header: checked_signature.content_sha256_header,
+	})
+}
+
+pub fn signing_hmac(
+	datetime: &DateTime<Utc>,
+	secret_key: &str,
+	region: &str,
+	service: &str,
+) -> Result<HmacSha256, crypto_common::InvalidLength> {
+	let secret = String::from("AWS4") + secret_key;
+	let mut date_hmac = HmacSha256::new_from_slice(secret.as_bytes())?;
+	date_hmac.update(datetime.format(SHORT_DATE).to_string().as_bytes());
+	let mut region_hmac = HmacSha256::new_from_slice(&date_hmac.finalize().into_bytes())?;
+	region_hmac.update(region.as_bytes());
+	let mut service_hmac = HmacSha256::new_from_slice(&region_hmac.finalize().into_bytes())?;
+	service_hmac.update(service.as_bytes());
+	let mut signing_hmac = HmacSha256::new_from_slice(&service_hmac.finalize().into_bytes())?;
+	signing_hmac.update(b"aws4_request");
+	let hmac = HmacSha256::new_from_slice(&signing_hmac.finalize().into_bytes())?;
+	Ok(hmac)
+}
+
+pub fn compute_scope(datetime: &DateTime<Utc>, region: &str, service: &str) -> String {
+	format!(
+		"{}/{}/{}/aws4_request",
+		datetime.format(SHORT_DATE),
+		region,
+		service
+	)
+}

src/api/common/signature/payload.rs

@@ -13,23 +13,9 @@ use garage_util::data::Hash;
 use garage_model::garage::Garage;
 use garage_model::key_table::*;
 
-use super::LONG_DATETIME;
+use super::*;
-use super::{compute_scope, signing_hmac};
 
 use crate::encoding::uri_encode;
-use crate::signature::error::*;
-
-pub const X_AMZ_ALGORITHM: HeaderName = HeaderName::from_static("x-amz-algorithm");
-pub const X_AMZ_CREDENTIAL: HeaderName = HeaderName::from_static("x-amz-credential");
-pub const X_AMZ_DATE: HeaderName = HeaderName::from_static("x-amz-date");
-pub const X_AMZ_EXPIRES: HeaderName = HeaderName::from_static("x-amz-expires");
-pub const X_AMZ_SIGNEDHEADERS: HeaderName = HeaderName::from_static("x-amz-signedheaders");
-pub const X_AMZ_SIGNATURE: HeaderName = HeaderName::from_static("x-amz-signature");
-pub const X_AMZ_CONTENT_SH256: HeaderName = HeaderName::from_static("x-amz-content-sha256");
-
-pub const AWS4_HMAC_SHA256: &str = "AWS4-HMAC-SHA256";
-pub const UNSIGNED_PAYLOAD: &str = "UNSIGNED-PAYLOAD";
-pub const STREAMING_AWS4_HMAC_SHA256_PAYLOAD: &str = "STREAMING-AWS4-HMAC-SHA256-PAYLOAD";
 
 pub type QueryMap = HeaderMap<QueryValue>;
 pub struct QueryValue {
@@ -39,16 +25,23 @@ pub struct QueryValue {
 	value: String,
 }
 
+#[derive(Debug)]
+pub struct CheckedSignature {
+	pub key: Option<Key>,
+	pub content_sha256_header: ContentSha256Header,
+	pub signature_header: Option<String>,
+}
+
 pub async fn check_payload_signature(
 	garage: &Garage,
 	request: &mut Request<IncomingBody>,
 	service: &'static str,
-) -> Result<(Option<Key>, Option<Hash>), Error> {
+) -> Result<CheckedSignature, Error> {
 	let query = parse_query_map(request.uri())?;
 
 	if query.contains_key(&X_AMZ_ALGORITHM) {
-		// We check for presigned-URL-style authentification first, because
+		// We check for presigned-URL-style authentication first, because
-		// the browser or someting else could inject an Authorization header
+		// the browser or something else could inject an Authorization header
 		// that is totally unrelated to AWS signatures.
 		check_presigned_signature(garage, service, request, query).await
 	} else if request.headers().contains_key(AUTHORIZATION) {
@@ -57,17 +50,46 @@ pub async fn check_payload_signature(
 		// Unsigned (anonymous) request
 		let content_sha256 = request
 			.headers()
-			.get("x-amz-content-sha256")
+			.get(X_AMZ_CONTENT_SHA256)
-			.filter(|c| c.as_bytes() != UNSIGNED_PAYLOAD.as_bytes());
+			.map(|x| x.to_str())
-		if let Some(content_sha256) = content_sha256 {
+			.transpose()?;
-			let sha256 = hex::decode(content_sha256)
+		Ok(CheckedSignature {
-				.ok()
+			key: None,
-				.and_then(|bytes| Hash::try_from(&bytes))
+			content_sha256_header: parse_x_amz_content_sha256(content_sha256)?,
-				.ok_or_bad_request("Invalid content sha256 hash")?;
+			signature_header: None,
-			Ok((None, Some(sha256)))
+		})
+	}
+}
+
+fn parse_x_amz_content_sha256(header: Option<&str>) -> Result<ContentSha256Header, Error> {
+	let header = match header {
+		Some(x) => x,
+		None => return Ok(ContentSha256Header::UnsignedPayload),
+	};
+	if header == UNSIGNED_PAYLOAD {
+		Ok(ContentSha256Header::UnsignedPayload)
+	} else if let Some(rest) = header.strip_prefix("STREAMING-") {
+		let (trailer, algo) = if let Some(rest2) = rest.strip_suffix("-TRAILER") {
+			(true, rest2)
 		} else {
-			Ok((None, None))
+			(false, rest)
-		}
+		};
+		let signed = match algo {
+			AWS4_HMAC_SHA256_PAYLOAD => true,
+			UNSIGNED_PAYLOAD => false,
+			_ => {
+				return Err(Error::bad_request(
+					"invalid or unsupported x-amz-content-sha256",
+				))
+			}
+		};
+		Ok(ContentSha256Header::StreamingPayload { trailer, signed })
+	} else {
+		let sha256 = hex::decode(header)
+			.ok()
+			.and_then(|bytes| Hash::try_from(&bytes))
+			.ok_or_bad_request("Invalid content sha256 hash")?;
+		Ok(ContentSha256Header::Sha256Checksum(sha256))
 	}
 }
 
@@ -76,13 +98,13 @@ async fn check_standard_signature(
 	service: &'static str,
 	request: &Request<IncomingBody>,
 	query: QueryMap,
-) -> Result<(Option<Key>, Option<Hash>), Error> {
+) -> Result<CheckedSignature, Error> {
 	let authorization = Authorization::parse_header(request.headers())?;
 
 	// Verify that all necessary request headers are included in signed_headers
 	// The following must be included for all signatures:
 	// - the Host header (mandatory)
-	// - all x-amz-* headers used in the request
+	// - all x-amz-* headers used in the request (except x-amz-content-sha256)
 	// AWS also indicates that the Content-Type header should be signed if
 	// it is used, but Minio client doesn't sign it so we don't check it for compatibility.
 	let signed_headers = split_signed_headers(&authorization)?;
@@ -108,18 +130,13 @@ async fn check_standard_signature(
 
 	let key = verify_v4(garage, service, &authorization, string_to_sign.as_bytes()).await?;
 
-	let content_sha256 = if authorization.content_sha256 == UNSIGNED_PAYLOAD {
+	let content_sha256_header = parse_x_amz_content_sha256(Some(&authorization.content_sha256))?;
-		None
-	} else if authorization.content_sha256 == STREAMING_AWS4_HMAC_SHA256_PAYLOAD {
-		let bytes = hex::decode(authorization.signature).ok_or_bad_request("Invalid signature")?;
-		Some(Hash::try_from(&bytes).ok_or_bad_request("Invalid signature")?)
-	} else {
-		let bytes = hex::decode(authorization.content_sha256)
-			.ok_or_bad_request("Invalid content sha256 hash")?;
-		Some(Hash::try_from(&bytes).ok_or_bad_request("Invalid content sha256 hash")?)
-	};
 
-	Ok((Some(key), content_sha256))
+	Ok(CheckedSignature {
+		key: Some(key),
+		content_sha256_header,
+		signature_header: Some(authorization.signature),
+	})
 }
 
 async fn check_presigned_signature(
@@ -127,14 +144,14 @@ async fn check_presigned_signature(
 	service: &'static str,
 	request: &mut Request<IncomingBody>,
 	mut query: QueryMap,
-) -> Result<(Option<Key>, Option<Hash>), Error> {
+) -> Result<CheckedSignature, Error> {
 	let algorithm = query.get(&X_AMZ_ALGORITHM).unwrap();
 	let authorization = Authorization::parse_presigned(&algorithm.value, &query)?;
 
 	// Verify that all necessary request headers are included in signed_headers
-	// For AWSv4 pre-signed URLs, the following must be incldued:
+	// For AWSv4 pre-signed URLs, the following must be included:
 	// - the Host header (mandatory)
-	// - all x-amz-* headers used in the request
+	// - all x-amz-* headers used in the request (except x-amz-content-sha256)
 	let signed_headers = split_signed_headers(&authorization)?;
 	verify_signed_headers(request.headers(), &signed_headers)?;
 
@@ -193,7 +210,11 @@ async fn check_presigned_signature(
 
 	// Presigned URLs always use UNSIGNED-PAYLOAD,
 	// so there is no sha256 hash to return.
-	Ok((Some(key), None))
+	Ok(CheckedSignature {
+		key: Some(key),
+		content_sha256_header: ContentSha256Header::UnsignedPayload,
+		signature_header: Some(authorization.signature),
+	})
 }
 
 pub fn parse_query_map(uri: &http::uri::Uri) -> Result<QueryMap, Error> {
@@ -247,7 +268,9 @@ fn verify_signed_headers(headers: &HeaderMap, signed_headers: &[HeaderName]) ->
 		return Err(Error::bad_request("Header `Host` should be signed"));
 	}
 	for (name, _) in headers.iter() {
-		if name.as_str().starts_with("x-amz-") {
+		// Enforce signature of all x-amz-* headers, except x-amz-content-sh256
+		// because it is included in the canonical request in all cases
+		if name.as_str().starts_with("x-amz-") && name != X_AMZ_CONTENT_SHA256 {
 			if !signed_headers.contains(name) {
 				return Err(Error::bad_request(format!(
 					"Header `{}` should be signed",
@@ -306,7 +329,7 @@ pub fn canonical_request(
 	// Note that there is also the issue of path normalization, which I hope is unrelated to the
 	// one of URI-encoding. At least in aws-sigv4 both parameters can be set independently,
 	// and rusoto_signature does not seem to do any effective path normalization, even though
-	// it mentions it in the comments (same link to the souce code as above).
+	// it mentions it in the comments (same link to the source code as above).
 	// We make the explicit choice of NOT normalizing paths in the K2V API because doing so
 	// would make non-normalized paths invalid K2V partition keys, and we don't want that.
 	let canonical_uri: std::borrow::Cow<str> = if service != "s3" {
@@ -396,7 +419,7 @@ pub async fn verify_v4(
 // ============ Authorization header, or X-Amz-* query params =========
 
 pub struct Authorization {
-	key_id: String,
+	pub key_id: String,
 	scope: String,
 	signed_headers: String,
 	signature: String,
@@ -405,7 +428,7 @@ pub struct Authorization {
 }
 
 impl Authorization {
-	fn parse_header(headers: &HeaderMap) -> Result<Self, Error> {
+	pub fn parse_header(headers: &HeaderMap) -> Result<Self, Error> {
 		let authorization = headers
 			.get(AUTHORIZATION)
 			.ok_or_bad_request("Missing authorization header")?
@@ -442,13 +465,12 @@ impl Authorization {
 			.to_string();
 
 		let content_sha256 = headers
-			.get(X_AMZ_CONTENT_SH256)
+			.get(X_AMZ_CONTENT_SHA256)
 			.ok_or_bad_request("Missing X-Amz-Content-Sha256 field")?;
 
 		let date = headers
 			.get(X_AMZ_DATE)
-			.ok_or_bad_request("Missing X-Amz-Date field")
+			.ok_or_bad_request("Missing X-Amz-Date field")?
-			.map_err(Error::from)?
 			.to_str()?;
 		let date = parse_date(date)?;
 
@@ -518,7 +540,7 @@ impl Authorization {
 		})
 	}
 
-	pub(crate) fn parse_form(params: &HeaderMap) -> Result<Self, Error> {
+	pub fn parse_form(params: &HeaderMap) -> Result<Self, Error> {
 		let algorithm = params
 			.get(X_AMZ_ALGORITHM)
 			.ok_or_bad_request("Missing X-Amz-Algorithm header")?

src/api/common/signature/streaming.rs

@@ -0,0 +1,618 @@
+use std::pin::Pin;
+use std::sync::Mutex;
+
+use chrono::{DateTime, NaiveDateTime, TimeZone, Utc};
+use futures::prelude::*;
+use futures::task;
+use hmac::Mac;
+use http::header::{HeaderMap, HeaderValue, CONTENT_ENCODING};
+use hyper::body::{Bytes, Frame, Incoming as IncomingBody};
+use hyper::Request;
+
+use garage_util::data::Hash;
+
+use super::*;
+
+use crate::helpers::body_stream;
+use crate::signature::checksum::*;
+use crate::signature::payload::CheckedSignature;
+
+pub use crate::signature::body::ReqBody;
+
+pub fn parse_streaming_body(
+	mut req: Request<IncomingBody>,
+	checked_signature: &CheckedSignature,
+	region: &str,
+	service: &str,
+) -> Result<Request<ReqBody>, Error> {
+	debug!(
+		"Content signature mode: {:?}",
+		checked_signature.content_sha256_header
+	);
+
+	match checked_signature.content_sha256_header {
+		ContentSha256Header::StreamingPayload { signed, trailer } => {
+			// Sanity checks
+			if !signed && !trailer {
+				return Err(Error::bad_request(
+					"STREAMING-UNSIGNED-PAYLOAD without trailer is not a valid combination",
+				));
+			}
+
+			// Remove the aws-chunked component in the content-encoding: header
+			// Note: this header is not properly sent by minio client, so don't fail
+			// if it is absent from the request.
+			if let Some(content_encoding) = req.headers_mut().remove(CONTENT_ENCODING) {
+				if let Some(rest) = content_encoding.as_bytes().strip_prefix(b"aws-chunked,") {
+					req.headers_mut()
+						.insert(CONTENT_ENCODING, HeaderValue::from_bytes(rest).unwrap());
+				} else if content_encoding != "aws-chunked" {
+					return Err(Error::bad_request(
+						"content-encoding does not contain aws-chunked for STREAMING-*-PAYLOAD",
+					));
+				}
+			}
+
+			// If trailer header is announced, add the calculation of the requested checksum
+			let mut checksummer = Checksummer::init(&Default::default(), false);
+			let trailer_algorithm = if trailer {
+				let algo = Some(
+					request_trailer_checksum_algorithm(req.headers())?
+						.ok_or_bad_request("Missing x-amz-trailer header")?,
+				);
+				checksummer = checksummer.add(algo);
+				algo
+			} else {
+				None
+			};
+
+			// For signed variants, determine signing parameters
+			let sign_params = if signed {
+				let signature = checked_signature
+					.signature_header
+					.clone()
+					.ok_or_bad_request("No signature provided")?;
+				let signature = hex::decode(signature)
+					.ok()
+					.and_then(|bytes| Hash::try_from(&bytes))
+					.ok_or_bad_request("Invalid signature")?;
+
+				let secret_key = checked_signature
+					.key
+					.as_ref()
+					.ok_or_bad_request("Cannot sign streaming payload without signing key")?
+					.state
+					.as_option()
+					.ok_or_internal_error("Deleted key state")?
+					.secret_key
+					.to_string();
+
+				let date = req
+					.headers()
+					.get(X_AMZ_DATE)
+					.ok_or_bad_request("Missing X-Amz-Date field")?
+					.to_str()?;
+				let date: NaiveDateTime = NaiveDateTime::parse_from_str(date, LONG_DATETIME)
+					.ok_or_bad_request("Invalid date")?;
+				let date: DateTime<Utc> = Utc.from_utc_datetime(&date);
+
+				let scope = compute_scope(&date, region, service);
+				let signing_hmac =
+					crate::signature::signing_hmac(&date, &secret_key, region, service)
+						.ok_or_internal_error("Unable to build signing HMAC")?;
+
+				Some(SignParams {
+					datetime: date,
+					scope,
+					signing_hmac,
+					previous_signature: signature,
+				})
+			} else {
+				None
+			};
+
+			Ok(req.map(move |body| {
+				let stream = body_stream::<_, Error>(body);
+
+				let signed_payload_stream =
+					StreamingPayloadStream::new(stream, sign_params, trailer).map_err(Error::from);
+				ReqBody {
+					stream: Mutex::new(signed_payload_stream.boxed()),
+					checksummer,
+					expected_checksums: Default::default(),
+					trailer_algorithm,
+				}
+			}))
+		}
+		_ => Ok(req.map(|body| {
+			let expected_checksums = ExpectedChecksums {
+				sha256: match &checked_signature.content_sha256_header {
+					ContentSha256Header::Sha256Checksum(sha256) => Some(*sha256),
+					_ => None,
+				},
+				..Default::default()
+			};
+			let checksummer = Checksummer::init(&expected_checksums, false);
+
+			let stream = http_body_util::BodyStream::new(body).map_err(Error::from);
+			ReqBody {
+				stream: Mutex::new(stream.boxed()),
+				checksummer,
+				expected_checksums,
+				trailer_algorithm: None,
+			}
+		})),
+	}
+}
+
+fn compute_streaming_payload_signature(
+	signing_hmac: &HmacSha256,
+	date: DateTime<Utc>,
+	scope: &str,
+	previous_signature: Hash,
+	content_sha256: Hash,
+) -> Result<Hash, StreamingPayloadError> {
+	let string_to_sign = [
+		AWS4_HMAC_SHA256_PAYLOAD,
+		&date.format(LONG_DATETIME).to_string(),
+		scope,
+		&hex::encode(previous_signature),
+		EMPTY_STRING_HEX_DIGEST,
+		&hex::encode(content_sha256),
+	]
+	.join("\n");
+
+	let mut hmac = signing_hmac.clone();
+	hmac.update(string_to_sign.as_bytes());
+
+	Hash::try_from(&hmac.finalize().into_bytes())
+		.ok_or_else(|| StreamingPayloadError::Message("Could not build signature".into()))
+}
+
+fn compute_streaming_trailer_signature(
+	signing_hmac: &HmacSha256,
+	date: DateTime<Utc>,
+	scope: &str,
+	previous_signature: Hash,
+	trailer_sha256: Hash,
+) -> Result<Hash, StreamingPayloadError> {
+	let string_to_sign = [
+		AWS4_HMAC_SHA256_PAYLOAD,
+		&date.format(LONG_DATETIME).to_string(),
+		scope,
+		&hex::encode(previous_signature),
+		&hex::encode(trailer_sha256),
+	]
+	.join("\n");
+
+	let mut hmac = signing_hmac.clone();
+	hmac.update(string_to_sign.as_bytes());
+
+	Hash::try_from(&hmac.finalize().into_bytes())
+		.ok_or_else(|| StreamingPayloadError::Message("Could not build signature".into()))
+}
+
+mod payload {
+	use http::{HeaderName, HeaderValue};
+
+	use garage_util::data::Hash;
+
+	use nom::bytes::streaming::{tag, take_while};
+	use nom::character::streaming::hex_digit1;
+	use nom::combinator::{map_res, opt};
+	use nom::number::streaming::hex_u32;
+
+	macro_rules! try_parse {
+		($expr:expr) => {
+			$expr.map_err(|e| e.map(Error::Parser))?
+		};
+	}
+
+	pub enum Error<I> {
+		Parser(nom::error::Error<I>),
+		BadSignature,
+	}
+
+	impl<I> Error<I> {
+		pub fn description(&self) -> &str {
+			match *self {
+				Error::Parser(ref e) => e.code.description(),
+				Error::BadSignature => "Bad signature",
+			}
+		}
+	}
+
+	#[derive(Debug, Clone)]
+	pub struct ChunkHeader {
+		pub size: usize,
+		pub signature: Option<Hash>,
+	}
+
+	impl ChunkHeader {
+		pub fn parse_signed(input: &[u8]) -> nom::IResult<&[u8], Self, Error<&[u8]>> {
+			let (input, size) = try_parse!(hex_u32(input));
+			let (input, _) = try_parse!(tag(";")(input));
+
+			let (input, _) = try_parse!(tag("chunk-signature=")(input));
+			let (input, data) = try_parse!(map_res(hex_digit1, hex::decode)(input));
+			let signature = Hash::try_from(&data).ok_or(nom::Err::Failure(Error::BadSignature))?;
+
+			let (input, _) = try_parse!(tag("\r\n")(input));
+
+			let header = ChunkHeader {
+				size: size as usize,
+				signature: Some(signature),
+			};
+
+			Ok((input, header))
+		}
+
+		pub fn parse_unsigned(input: &[u8]) -> nom::IResult<&[u8], Self, Error<&[u8]>> {
+			let (input, size) = try_parse!(hex_u32(input));
+			let (input, _) = try_parse!(tag("\r\n")(input));
+
+			let header = ChunkHeader {
+				size: size as usize,
+				signature: None,
+			};
+
+			Ok((input, header))
+		}
+	}
+
+	#[derive(Debug, Clone)]
+	pub struct TrailerChunk {
+		pub header_name: HeaderName,
+		pub header_value: HeaderValue,
+		pub signature: Option<Hash>,
+	}
+
+	impl TrailerChunk {
+		fn parse_content(input: &[u8]) -> nom::IResult<&[u8], Self, Error<&[u8]>> {
+			let (input, header_name) = try_parse!(map_res(
+				take_while(|c: u8| c.is_ascii_alphanumeric() || c == b'-'),
+				HeaderName::from_bytes
+			)(input));
+			let (input, _) = try_parse!(tag(b":")(input));
+			let (input, header_value) = try_parse!(map_res(
+				take_while(|c: u8| c.is_ascii_alphanumeric() || b"+/=".contains(&c)),
+				HeaderValue::from_bytes
+			)(input));
+
+			// Possible '\n' after the header value, depends on clients
+			// https://docs.aws.amazon.com/AmazonS3/latest/userguide/checking-object-integrity.html
+			let (input, _) = try_parse!(opt(tag(b"\n"))(input));
+
+			let (input, _) = try_parse!(tag(b"\r\n")(input));
+
+			Ok((
+				input,
+				TrailerChunk {
+					header_name,
+					header_value,
+					signature: None,
+				},
+			))
+		}
+		pub fn parse_signed(input: &[u8]) -> nom::IResult<&[u8], Self, Error<&[u8]>> {
+			let (input, trailer) = Self::parse_content(input)?;
+
+			let (input, _) = try_parse!(tag(b"x-amz-trailer-signature:")(input));
+			let (input, data) = try_parse!(map_res(hex_digit1, hex::decode)(input));
+			let signature = Hash::try_from(&data).ok_or(nom::Err::Failure(Error::BadSignature))?;
+			let (input, _) = try_parse!(tag(b"\r\n")(input));
+
+			Ok((
+				input,
+				TrailerChunk {
+					signature: Some(signature),
+					..trailer
+				},
+			))
+		}
+		pub fn parse_unsigned(input: &[u8]) -> nom::IResult<&[u8], Self, Error<&[u8]>> {
+			let (input, trailer) = Self::parse_content(input)?;
+			let (input, _) = try_parse!(tag(b"\r\n")(input));
+
+			Ok((input, trailer))
+		}
+	}
+}
+
+#[derive(Debug)]
+pub enum StreamingPayloadError {
+	Stream(Error),
+	InvalidSignature,
+	Message(String),
+}
+
+impl StreamingPayloadError {
+	fn message(msg: &str) -> Self {
+		StreamingPayloadError::Message(msg.into())
+	}
+}
+
+impl From<StreamingPayloadError> for Error {
+	fn from(err: StreamingPayloadError) -> Self {
+		match err {
+			StreamingPayloadError::Stream(e) => e,
+			StreamingPayloadError::InvalidSignature => {
+				Error::bad_request("Invalid payload signature")
+			}
+			StreamingPayloadError::Message(e) => {
+				Error::bad_request(format!("Chunk format error: {}", e))
+			}
+		}
+	}
+}
+
+impl<I> From<payload::Error<I>> for StreamingPayloadError {
+	fn from(err: payload::Error<I>) -> Self {
+		Self::message(err.description())
+	}
+}
+
+impl<I> From<nom::error::Error<I>> for StreamingPayloadError {
+	fn from(err: nom::error::Error<I>) -> Self {
+		Self::message(err.code.description())
+	}
+}
+
+enum StreamingPayloadChunk {
+	Chunk {
+		header: payload::ChunkHeader,
+		data: Bytes,
+	},
+	Trailer(payload::TrailerChunk),
+}
+
+struct SignParams {
+	datetime: DateTime<Utc>,
+	scope: String,
+	signing_hmac: HmacSha256,
+	previous_signature: Hash,
+}
+
+#[pin_project::pin_project]
+pub struct StreamingPayloadStream<S>
+where
+	S: Stream<Item = Result<Bytes, Error>>,
+{
+	#[pin]
+	stream: S,
+	buf: bytes::BytesMut,
+	signing: Option<SignParams>,
+	has_trailer: bool,
+	done: bool,
+}
+
+impl<S> StreamingPayloadStream<S>
+where
+	S: Stream<Item = Result<Bytes, Error>>,
+{
+	fn new(stream: S, signing: Option<SignParams>, has_trailer: bool) -> Self {
+		Self {
+			stream,
+			buf: bytes::BytesMut::new(),
+			signing,
+			has_trailer,
+			done: false,
+		}
+	}
+
+	fn parse_next(
+		input: &[u8],
+		is_signed: bool,
+		has_trailer: bool,
+	) -> nom::IResult<&[u8], StreamingPayloadChunk, StreamingPayloadError> {
+		use nom::bytes::streaming::{tag, take};
+
+		macro_rules! try_parse {
+			($expr:expr) => {
+				$expr.map_err(nom::Err::convert)?
+			};
+		}
+
+		let (input, header) = if is_signed {
+			try_parse!(payload::ChunkHeader::parse_signed(input))
+		} else {
+			try_parse!(payload::ChunkHeader::parse_unsigned(input))
+		};
+
+		// 0-sized chunk is the last
+		if header.size == 0 {
+			if has_trailer {
+				let (input, trailer) = if is_signed {
+					try_parse!(payload::TrailerChunk::parse_signed(input))
+				} else {
+					try_parse!(payload::TrailerChunk::parse_unsigned(input))
+				};
+				return Ok((input, StreamingPayloadChunk::Trailer(trailer)));
+			} else {
+				return Ok((
+					input,
+					StreamingPayloadChunk::Chunk {
+						header,
+						data: Bytes::new(),
+					},
+				));
+			}
+		}
+
+		let (input, data) = try_parse!(take::<_, _, nom::error::Error<_>>(header.size)(input));
+		let (input, _) = try_parse!(tag::<_, _, nom::error::Error<_>>("\r\n")(input));
+
+		let data = Bytes::from(data.to_vec());
+
+		Ok((input, StreamingPayloadChunk::Chunk { header, data }))
+	}
+}
+
+impl<S> Stream for StreamingPayloadStream<S>
+where
+	S: Stream<Item = Result<Bytes, Error>> + Unpin,
+{
+	type Item = Result<Frame<Bytes>, StreamingPayloadError>;
+
+	fn poll_next(
+		self: Pin<&mut Self>,
+		cx: &mut task::Context<'_>,
+	) -> task::Poll<Option<Self::Item>> {
+		use std::task::Poll;
+
+		let mut this = self.project();
+
+		if *this.done {
+			return Poll::Ready(None);
+		}
+
+		loop {
+			let (input, payload) =
+				match Self::parse_next(this.buf, this.signing.is_some(), *this.has_trailer) {
+					Ok(res) => res,
+					Err(nom::Err::Incomplete(_)) => {
+						match futures::ready!(this.stream.as_mut().poll_next(cx)) {
+							Some(Ok(bytes)) => {
+								this.buf.extend(bytes);
+								continue;
+							}
+							Some(Err(e)) => {
+								return Poll::Ready(Some(Err(StreamingPayloadError::Stream(e))))
+							}
+							None => {
+								return Poll::Ready(Some(Err(StreamingPayloadError::message(
+									"Unexpected EOF",
+								))));
+							}
+						}
+					}
+					Err(nom::Err::Error(e)) | Err(nom::Err::Failure(e)) => {
+						return Poll::Ready(Some(Err(e)))
+					}
+				};
+
+			match payload {
+				StreamingPayloadChunk::Chunk { data, header } => {
+					if let Some(signing) = this.signing.as_mut() {
+						let data_sha256sum = sha256sum(&data);
+
+						let expected_signature = compute_streaming_payload_signature(
+							&signing.signing_hmac,
+							signing.datetime,
+							&signing.scope,
+							signing.previous_signature,
+							data_sha256sum,
+						)?;
+
+						if header.signature.unwrap() != expected_signature {
+							return Poll::Ready(Some(Err(StreamingPayloadError::InvalidSignature)));
+						}
+
+						signing.previous_signature = header.signature.unwrap();
+					}
+
+					*this.buf = input.into();
+
+					// 0-sized chunk is the last
+					if data.is_empty() {
+						// if there was a trailer, it would have been returned by the parser
+						assert!(!*this.has_trailer);
+						*this.done = true;
+						return Poll::Ready(None);
+					}
+
+					return Poll::Ready(Some(Ok(Frame::data(data))));
+				}
+				StreamingPayloadChunk::Trailer(trailer) => {
+					trace!(
+						"In StreamingPayloadStream::poll_next: got trailer {:?}",
+						trailer
+					);
+
+					if let Some(signing) = this.signing.as_mut() {
+						let data = [
+							trailer.header_name.as_ref(),
+							&b":"[..],
+							trailer.header_value.as_ref(),
+							&b"\n"[..],
+						]
+						.concat();
+						let trailer_sha256sum = sha256sum(&data);
+
+						let expected_signature = compute_streaming_trailer_signature(
+							&signing.signing_hmac,
+							signing.datetime,
+							&signing.scope,
+							signing.previous_signature,
+							trailer_sha256sum,
+						)?;
+
+						if trailer.signature.unwrap() != expected_signature {
+							return Poll::Ready(Some(Err(StreamingPayloadError::InvalidSignature)));
+						}
+					}
+
+					*this.buf = input.into();
+					*this.done = true;
+
+					let mut trailers_map = HeaderMap::new();
+					trailers_map.insert(trailer.header_name, trailer.header_value);
+
+					return Poll::Ready(Some(Ok(Frame::trailers(trailers_map))));
+				}
+			}
+		}
+	}
+
+	fn size_hint(&self) -> (usize, Option<usize>) {
+		self.stream.size_hint()
+	}
+}
+
+#[cfg(test)]
+mod tests {
+	use futures::prelude::*;
+
+	use super::{SignParams, StreamingPayloadError, StreamingPayloadStream};
+
+	#[tokio::test]
+	async fn test_interrupted_signed_payload_stream() {
+		use chrono::{DateTime, Utc};
+
+		use garage_util::data::Hash;
+
+		let datetime = DateTime::parse_from_rfc3339("2021-12-13T13:12:42+01:00") // TODO UNIX 0
+			.unwrap()
+			.with_timezone(&Utc);
+		let secret_key = "test";
+		let region = "test";
+		let scope = crate::signature::compute_scope(&datetime, region, "s3");
+		let signing_hmac =
+			crate::signature::signing_hmac(&datetime, secret_key, region, "s3").unwrap();
+
+		let data: &[&[u8]] = &[b"1"];
+		let body = futures::stream::iter(data.iter().map(|block| Ok(block.to_vec().into())));
+
+		let seed_signature = Hash::default();
+
+		let mut stream = StreamingPayloadStream::new(
+			body,
+			Some(SignParams {
+				signing_hmac,
+				datetime,
+				scope,
+				previous_signature: seed_signature,
+			}),
+			false,
+		);
+
+		assert!(stream.try_next().await.is_err());
+		match stream.try_next().await {
+			Err(StreamingPayloadError::Message(msg)) if msg == "Unexpected EOF" => {}
+			item => panic!(
+				"Unexpected result, expected early EOF error, got {:?}",
+				item
+			),
+		}
+	}
+}

src/api/k2v/Cargo.toml

@@ -0,0 +1,37 @@
+[package]
+name = "garage_api_k2v"
+version = "1.3.1"
+authors = ["Alex Auvolat <alex@adnab.me>"]
+edition = "2018"
+license = "AGPL-3.0"
+description = "K2V API server crate for the Garage object store"
+repository = "https://git.deuxfleurs.fr/Deuxfleurs/garage"
+readme = "../../../README.md"
+
+[lib]
+path = "lib.rs"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+garage_model = { workspace = true, features = [ "k2v" ] }
+garage_table.workspace = true
+garage_util = { workspace = true, features = [ "k2v" ] }
+garage_api_common.workspace = true
+
+base64.workspace = true
+thiserror.workspace = true
+tracing.workspace = true
+
+futures.workspace = true
+tokio.workspace = true
+http.workspace = true
+http-body-util.workspace = true
+hyper = { workspace = true, default-features = false, features = ["server", "http1"] }
+percent-encoding.workspace = true
+url.workspace = true
+
+serde.workspace = true
+serde_json.workspace = true
+
+opentelemetry.workspace = true

src/api/k2v/api_server.rs

@@ -1,7 +1,5 @@
 use std::sync::Arc;
 
-use async_trait::async_trait;
-
 use hyper::{body::Incoming as IncomingBody, Method, Request, Response};
 use tokio::sync::watch;
 
@@ -12,26 +10,25 @@ use garage_util::socket_address::UnixOrTCPSocketAddress;
 
 use garage_model::garage::Garage;
 
-use crate::generic_server::*;
+use garage_api_common::cors::*;
-use crate::k2v::error::*;
+use garage_api_common::generic_server::*;
+use garage_api_common::helpers::*;
+use garage_api_common::signature::verify_request;
 
-use crate::signature::verify_request;
+use crate::batch::*;
+use crate::error::*;
+use crate::index::*;
+use crate::item::*;
+use crate::router::Endpoint;
 
-use crate::helpers::*;
+pub use garage_api_common::signature::streaming::ReqBody;
-use crate::k2v::batch::*;
-use crate::k2v::index::*;
-use crate::k2v::item::*;
-use crate::k2v::router::Endpoint;
-use crate::s3::cors::*;
-
-pub use crate::signature::streaming::ReqBody;
 pub type ResBody = BoxBody<Error>;
 
 pub struct K2VApiServer {
 	garage: Arc<Garage>,
 }
 
-pub(crate) struct K2VApiEndpoint {
+pub struct K2VApiEndpoint {
 	bucket_name: String,
 	endpoint: Endpoint,
 }
@@ -49,7 +46,6 @@ impl K2VApiServer {
 	}
 }
 
-#[async_trait]
 impl ApiHandler for K2VApiServer {
 	const API_NAME: &'static str = "k2v";
 	const API_NAME_DISPLAY: &'static str = "K2V";
@@ -77,7 +73,7 @@ impl ApiHandler for K2VApiServer {
 		} = endpoint;
 		let garage = self.garage.clone();
 
-		// The OPTIONS method is procesed early, before we even check for an API key
+		// The OPTIONS method is processed early, before we even check for an API key
 		if let Endpoint::Options = endpoint {
 			let options_res = handle_options_api(garage, &req, Some(bucket_name))
 				.await
@@ -85,16 +81,20 @@ impl ApiHandler for K2VApiServer {
 			return Ok(options_res.map(|_empty_body: EmptyBody| empty_body()));
 		}
 
-		let (req, api_key, _content_sha256) = verify_request(&garage, req, "k2v").await?;
+		let verified_request = verify_request(&garage, req, "k2v").await?;
+		let req = verified_request.request;
+		let api_key = verified_request.access_key;
 
 		let bucket_id = garage
 			.bucket_helper()
 			.resolve_bucket(&bucket_name, &api_key)
-			.await?;
+			.await
+			.map_err(pass_helper_error)?;
 		let bucket = garage
 			.bucket_helper()
 			.get_existing_bucket(bucket_id)
-			.await?;
+			.await
+			.map_err(helper_error_as_internal)?;
 		let bucket_params = bucket.state.into_option().unwrap();
 
 		let allowed = match endpoint.authorization_type() {
@@ -176,6 +176,12 @@ impl ApiHandler for K2VApiServer {
 
 		Ok(resp_ok)
 	}
+
+	fn key_id_from_request(&self, req: &Request<IncomingBody>) -> Option<String> {
+		garage_api_common::signature::payload::Authorization::parse_header(req.headers())
+			.map(|auth| auth.key_id)
+			.ok()
+	}
 }
 
 impl ApiEndpoint for K2VApiEndpoint {

src/api/k2v/batch.rs

@@ -4,13 +4,14 @@ use serde::{Deserialize, Serialize};
 
 use garage_table::{EnumerationOrder, TableSchema};
 
-use garage_model::k2v::causality::*;
 use garage_model::k2v::item_table::*;
 
-use crate::helpers::*;
+use garage_api_common::helpers::*;
-use crate::k2v::api_server::{ReqBody, ResBody};
+
-use crate::k2v::error::*;
+use crate::api_server::{ReqBody, ResBody};
-use crate::k2v::range::read_range;
+use crate::error::*;
+use crate::item::parse_causality_token;
+use crate::range::read_range;
 
 pub async fn handle_insert_batch(
 	ctx: ReqCtx,
@@ -19,11 +20,11 @@ pub async fn handle_insert_batch(
 	let ReqCtx {
 		garage, bucket_id, ..
 	} = &ctx;
-	let items = parse_json_body::<Vec<InsertBatchItem>, _, Error>(req).await?;
+	let items = req.into_body().json::<Vec<InsertBatchItem>>().await?;
 
 	let mut items2 = vec![];
 	for it in items {
-		let ct = it.ct.map(|s| CausalContext::parse_helper(&s)).transpose()?;
+		let ct = it.ct.map(|s| parse_causality_token(&s)).transpose()?;
 		let v = match it.v {
 			Some(vs) => DvvsValue::Value(
 				BASE64_STANDARD
@@ -46,7 +47,7 @@ pub async fn handle_read_batch(
 	ctx: ReqCtx,
 	req: Request<ReqBody>,
 ) -> Result<Response<ResBody>, Error> {
-	let queries = parse_json_body::<Vec<ReadBatchQuery>, _, Error>(req).await?;
+	let queries = req.into_body().json::<Vec<ReadBatchQuery>>().await?;
 
 	let resp_results = futures::future::join_all(
 		queries
@@ -140,7 +141,7 @@ pub async fn handle_delete_batch(
 	ctx: ReqCtx,
 	req: Request<ReqBody>,
 ) -> Result<Response<ResBody>, Error> {
-	let queries = parse_json_body::<Vec<DeleteBatchQuery>, _, Error>(req).await?;
+	let queries = req.into_body().json::<Vec<DeleteBatchQuery>>().await?;
 
 	let resp_results = futures::future::join_all(
 		queries
@@ -261,7 +262,7 @@ pub(crate) async fn handle_poll_range(
 	} = ctx;
 	use garage_model::k2v::sub::PollRange;
 
-	let query = parse_json_body::<PollRangeQuery, _, Error>(req).await?;
+	let query = req.into_body().json::<PollRangeQuery>().await?;
 
 	let timeout_msec = query.timeout.unwrap_or(300).clamp(1, 600) * 1000;
 
@@ -281,7 +282,8 @@ pub(crate) async fn handle_poll_range(
 			query.seen_marker,
 			timeout_msec,
 		)
-		.await?;
+		.await
+		.map_err(pass_helper_error)?;
 
 	if let Some((items, seen_marker)) = resp {
 		let resp = PollRangeResponse {

src/api/k2v/error.rs

@@ -1,52 +1,54 @@
-use err_derive::Error;
 use hyper::header::HeaderValue;
 use hyper::{HeaderMap, StatusCode};
+use thiserror::Error;
 
-use crate::common_error::CommonError;
+use garage_api_common::common_error::{commonErrorDerivative, CommonError};
-pub use crate::common_error::{CommonErrorDerivative, OkOrBadRequest, OkOrInternalError};
+pub(crate) use garage_api_common::common_error::{helper_error_as_internal, pass_helper_error};
-use crate::generic_server::ApiError;
+pub use garage_api_common::common_error::{
-use crate::helpers::*;
+	CommonErrorDerivative, OkOrBadRequest, OkOrInternalError,
-use crate::signature::error::Error as SignatureError;
+};
+use garage_api_common::generic_server::ApiError;
+use garage_api_common::helpers::*;
+use garage_api_common::signature::error::Error as SignatureError;
 
 /// Errors of this crate
 #[derive(Debug, Error)]
 pub enum Error {
-	#[error(display = "{}", _0)]
+	#[error("{0}")]
 	/// Error from common error
-	Common(CommonError),
+	Common(#[from] CommonError),
 
 	// Category: cannot process
 	/// Authorization Header Malformed
-	#[error(display = "Authorization header malformed, unexpected scope: {}", _0)]
+	#[error("Authorization header malformed, unexpected scope: {0}")]
 	AuthorizationHeaderMalformed(String),
 
+	/// The provided digest (checksum) value was invalid
+	#[error("Invalid digest: {0}")]
+	InvalidDigest(String),
+
 	/// The object requested don't exists
-	#[error(display = "Key not found")]
+	#[error("Key not found")]
 	NoSuchKey,
 
 	/// Some base64 encoded data was badly encoded
-	#[error(display = "Invalid base64: {}", _0)]
+	#[error("Invalid base64: {0}")]
-	InvalidBase64(#[error(source)] base64::DecodeError),
+	InvalidBase64(#[from] base64::DecodeError),
+
+	/// Invalid causality token
+	#[error("Invalid causality token")]
+	InvalidCausalityToken,
 
 	/// The client asked for an invalid return format (invalid Accept header)
-	#[error(display = "Not acceptable: {}", _0)]
+	#[error("Not acceptable: {0}")]
 	NotAcceptable(String),
 
 	/// The request contained an invalid UTF-8 sequence in its path or in other parameters
-	#[error(display = "Invalid UTF-8: {}", _0)]
+	#[error("Invalid UTF-8: {0}")]
-	InvalidUtf8Str(#[error(source)] std::str::Utf8Error),
+	InvalidUtf8Str(#[from] std::str::Utf8Error),
 }
 
-impl<T> From<T> for Error
+commonErrorDerivative!(Error);
-where
-	CommonError: From<T>,
-{
-	fn from(err: T) -> Self {
-		Error::Common(CommonError::from(err))
-	}
-}
-
-impl CommonErrorDerivative for Error {}
 
 impl From<SignatureError> for Error {
 	fn from(err: SignatureError) -> Self {
@@ -56,6 +58,7 @@ impl From<SignatureError> for Error {
 				Self::AuthorizationHeaderMalformed(c)
 			}
 			SignatureError::InvalidUtf8Str(i) => Self::InvalidUtf8Str(i),
+			SignatureError::InvalidDigest(d) => Self::InvalidDigest(d),
 		}
 	}
 }
@@ -72,6 +75,8 @@ impl Error {
 			Error::AuthorizationHeaderMalformed(_) => "AuthorizationHeaderMalformed",
 			Error::InvalidBase64(_) => "InvalidBase64",
 			Error::InvalidUtf8Str(_) => "InvalidUtf8String",
+			Error::InvalidCausalityToken => "CausalityToken",
+			Error::InvalidDigest(_) => "InvalidDigest",
 		}
 	}
 }
@@ -85,7 +90,9 @@ impl ApiError for Error {
 			Error::NotAcceptable(_) => StatusCode::NOT_ACCEPTABLE,
 			Error::AuthorizationHeaderMalformed(_)
 			| Error::InvalidBase64(_)
-			| Error::InvalidUtf8Str(_) => StatusCode::BAD_REQUEST,
+			| Error::InvalidUtf8Str(_)
+			| Error::InvalidDigest(_)
+			| Error::InvalidCausalityToken => StatusCode::BAD_REQUEST,
 		}
 	}
 

src/api/k2v/index.rs

@@ -5,10 +5,11 @@ use garage_table::util::*;
 
 use garage_model::k2v::item_table::{BYTES, CONFLICTS, ENTRIES, VALUES};
 
-use crate::helpers::*;
+use garage_api_common::helpers::*;
-use crate::k2v::api_server::ResBody;
+
-use crate::k2v::error::*;
+use crate::api_server::ResBody;
-use crate::k2v::range::read_range;
+use crate::error::*;
+use crate::range::read_range;
 
 pub async fn handle_read_index(
 	ctx: ReqCtx,

src/api/k2v/item.rs

@@ -6,9 +6,10 @@ use hyper::{Request, Response, StatusCode};
 use garage_model::k2v::causality::*;
 use garage_model::k2v::item_table::*;
 
-use crate::helpers::*;
+use garage_api_common::helpers::*;
-use crate::k2v::api_server::{ReqBody, ResBody};
+
-use crate::k2v::error::*;
+use crate::api_server::{ReqBody, ResBody};
+use crate::error::*;
 
 pub const X_GARAGE_CAUSALITY_TOKEN: &str = "X-Garage-Causality-Token";
 
@@ -18,6 +19,10 @@ pub enum ReturnFormat {
 	Either,
 }
 
+pub(crate) fn parse_causality_token(s: &str) -> Result<CausalContext, Error> {
+	CausalContext::parse(s).ok_or(Error::InvalidCausalityToken)
+}
+
 impl ReturnFormat {
 	pub fn from(req: &Request<ReqBody>) -> Result<Self, Error> {
 		let accept = match req.headers().get(header::ACCEPT) {
@@ -136,12 +141,10 @@ pub async fn handle_insert_item(
 		.get(X_GARAGE_CAUSALITY_TOKEN)
 		.map(|s| s.to_str())
 		.transpose()?
-		.map(CausalContext::parse_helper)
+		.map(parse_causality_token)
 		.transpose()?;
 
-	let body = http_body_util::BodyExt::collect(req.into_body())
+	let body = req.into_body().collect().await?;
-		.await?
-		.to_bytes();
 
 	let value = DvvsValue::Value(body.to_vec());
 
@@ -176,7 +179,7 @@ pub async fn handle_delete_item(
 		.get(X_GARAGE_CAUSALITY_TOKEN)
 		.map(|s| s.to_str())
 		.transpose()?
-		.map(CausalContext::parse_helper)
+		.map(parse_causality_token)
 		.transpose()?;
 
 	let value = DvvsValue::Deleted;

src/api/k2v/lib.rs

@@ -1,3 +1,6 @@
+#[macro_use]
+extern crate tracing;
+
 pub mod api_server;
 mod error;
 mod router;

src/api/k2v/range.rs

@@ -7,8 +7,9 @@ use std::sync::Arc;
 use garage_table::replication::TableShardedReplication;
 use garage_table::*;
 
-use crate::helpers::key_after_prefix;
+use garage_api_common::helpers::key_after_prefix;
-use crate::k2v::error::*;
+
+use crate::error::*;
 
 /// Read range in a Garage table.
 /// Returns (entries, more?, nextStart)

src/api/k2v/router.rs

@@ -1,11 +1,11 @@
-use crate::k2v::error::*;
+use crate::error::*;
 
 use std::borrow::Cow;
 
 use hyper::{Method, Request};
 
-use crate::helpers::Authorization;
+use garage_api_common::helpers::Authorization;
-use crate::router_macros::{generateQueryParameters, router_match};
+use garage_api_common::router_macros::{generateQueryParameters, router_match};
 
 router_match! {@func
 

src/api/lib.rs

@@ -1,17 +0,0 @@
-//! Crate for serving a S3 compatible API
-#[macro_use]
-extern crate tracing;
-
-pub mod common_error;
-
-mod encoding;
-pub mod generic_server;
-pub mod helpers;
-mod router_macros;
-/// This mode is public only to help testing. Don't expect stability here
-pub mod signature;
-
-pub mod admin;
-#[cfg(feature = "k2v")]
-pub mod k2v;
-pub mod s3;

src/api/s3/Cargo.toml

@@ -1,12 +1,12 @@
 [package]
-name = "garage_api"
+name = "garage_api_s3"
-version = "1.0.0"
+version = "1.3.1"
 authors = ["Alex Auvolat <alex@adnab.me>"]
 edition = "2018"
 license = "AGPL-3.0"
 description = "S3 API server crate for the Garage object store"
 repository = "https://git.deuxfleurs.fr/Deuxfleurs/garage"
-readme = "../../README.md"
+readme = "../../../README.md"
 
 [lib]
 path = "lib.rs"
@@ -20,30 +20,24 @@ garage_block.workspace = true
 garage_net.workspace = true
 garage_util.workspace = true
 garage_rpc.workspace = true
+garage_api_common.workspace = true
 
 aes-gcm.workspace = true
-argon2.workspace = true
 async-compression.workspace = true
-async-trait.workspace = true
 base64.workspace = true
 bytes.workspace = true
 chrono.workspace = true
 crc32fast.workspace = true
 crc32c.workspace = true
-crypto-common.workspace = true
+thiserror.workspace = true
-err-derive.workspace = true
 hex.workspace = true
-hmac.workspace = true
-idna.workspace = true
 tracing.workspace = true
 md-5.workspace = true
-nom.workspace = true
 pin-project.workspace = true
 sha1.workspace = true
 sha2.workspace = true
 
 futures.workspace = true
-futures-util.workspace = true
 tokio.workspace = true
 tokio-stream.workspace = true
 tokio-util.workspace = true
@@ -54,21 +48,13 @@ httpdate.workspace = true
 http-range.workspace = true
 http-body-util.workspace = true
 hyper = { workspace = true, default-features = false, features = ["server", "http1"] }
-hyper-util.workspace = true
 multer.workspace = true
 percent-encoding.workspace = true
 roxmltree.workspace = true
 url.workspace = true
 
 serde.workspace = true
-serde_bytes.workspace = true
 serde_json.workspace = true
 quick-xml.workspace = true
 
 opentelemetry.workspace = true
-opentelemetry-prometheus = { workspace = true, optional = true }
-prometheus = { workspace = true, optional = true }
-
-[features]
-k2v = [ "garage_util/k2v", "garage_model/k2v" ]
-metrics = [ "opentelemetry-prometheus", "prometheus" ]

src/api/s3/api_server.rs

@@ -1,7 +1,5 @@
 use std::sync::Arc;
 
-use async_trait::async_trait;
-
 use hyper::header;
 use hyper::{body::Incoming as IncomingBody, Request, Response};
 use tokio::sync::watch;
@@ -14,33 +12,33 @@ use garage_util::socket_address::UnixOrTCPSocketAddress;
 use garage_model::garage::Garage;
 use garage_model::key_table::Key;
 
-use crate::generic_server::*;
+use garage_api_common::cors::*;
-use crate::s3::error::*;
+use garage_api_common::generic_server::*;
+use garage_api_common::helpers::*;
+use garage_api_common::signature::verify_request;
 
-use crate::signature::verify_request;
+use crate::bucket::*;
+use crate::copy::*;
+use crate::cors::*;
+use crate::delete::*;
+use crate::error::*;
+use crate::get::*;
+use crate::lifecycle::*;
+use crate::list::*;
+use crate::multipart::*;
+use crate::post_object::handle_post_object;
+use crate::put::*;
+use crate::router::Endpoint;
+use crate::website::*;
 
-use crate::helpers::*;
+pub use garage_api_common::signature::streaming::ReqBody;
-use crate::s3::bucket::*;
-use crate::s3::copy::*;
-use crate::s3::cors::*;
-use crate::s3::delete::*;
-use crate::s3::get::*;
-use crate::s3::lifecycle::*;
-use crate::s3::list::*;
-use crate::s3::multipart::*;
-use crate::s3::post_object::handle_post_object;
-use crate::s3::put::*;
-use crate::s3::router::Endpoint;
-use crate::s3::website::*;
-
-pub use crate::signature::streaming::ReqBody;
 pub type ResBody = BoxBody<Error>;
 
 pub struct S3ApiServer {
 	garage: Arc<Garage>,
 }
 
-pub(crate) struct S3ApiEndpoint {
+pub struct S3ApiEndpoint {
 	bucket_name: Option<String>,
 	endpoint: Endpoint,
 }
@@ -70,7 +68,6 @@ impl S3ApiServer {
 	}
 }
 
-#[async_trait]
 impl ApiHandler for S3ApiServer {
 	const API_NAME: &'static str = "s3";
 	const API_NAME_DISPLAY: &'static str = "S3";
@@ -124,7 +121,9 @@ impl ApiHandler for S3ApiServer {
 			return Ok(options_res.map(|_empty_body: EmptyBody| empty_body()));
 		}
 
-		let (req, api_key, content_sha256) = verify_request(&garage, req, "s3").await?;
+		let verified_request = verify_request(&garage, req, "s3").await?;
+		let req = verified_request.request;
+		let api_key = verified_request.access_key;
 
 		let bucket_name = match bucket_name {
 			None => {
@@ -137,20 +136,14 @@ impl ApiHandler for S3ApiServer {
 
 		// Special code path for CreateBucket API endpoint
 		if let Endpoint::CreateBucket {} = endpoint {
-			return handle_create_bucket(
+			return handle_create_bucket(&garage, req, &api_key.key_id, bucket_name).await;
-				&garage,
-				req,
-				content_sha256,
-				&api_key.key_id,
-				bucket_name,
-			)
-			.await;
 		}
 
 		let bucket_id = garage
 			.bucket_helper()
 			.resolve_bucket(&bucket_name, &api_key)
-			.await?;
+			.await
+			.map_err(pass_helper_error)?;
 		let bucket = garage
 			.bucket_helper()
 			.get_existing_bucket(bucket_id)
@@ -181,7 +174,7 @@ impl ApiHandler for S3ApiServer {
 		let resp = match endpoint {
 			Endpoint::HeadObject {
 				key, part_number, ..
-			} => handle_head(ctx, &req, &key, part_number).await,
+			} => handle_head(ctx, &req.map(|_| ()), &key, part_number).await,
 			Endpoint::GetObject {
 				key,
 				part_number,
@@ -201,20 +194,20 @@ impl ApiHandler for S3ApiServer {
 					response_content_type,
 					response_expires,
 				};
-				handle_get(ctx, &req, &key, part_number, overrides).await
+				handle_get(ctx, &req.map(|_| ()), &key, part_number, overrides).await
 			}
 			Endpoint::UploadPart {
 				key,
 				part_number,
 				upload_id,
-			} => handle_put_part(ctx, req, &key, part_number, &upload_id, content_sha256).await,
+			} => handle_put_part(ctx, req, &key, part_number, &upload_id).await,
 			Endpoint::CopyObject { key } => handle_copy(ctx, &req, &key).await,
 			Endpoint::UploadPartCopy {
 				key,
 				part_number,
 				upload_id,
 			} => handle_upload_part_copy(ctx, &req, &key, part_number, &upload_id).await,
-			Endpoint::PutObject { key } => handle_put(ctx, req, &key, content_sha256).await,
+			Endpoint::PutObject { key } => handle_put(ctx, req, &key).await,
 			Endpoint::AbortMultipartUpload { key, upload_id } => {
 				handle_abort_multipart_upload(ctx, &key, &upload_id).await
 			}
@@ -223,7 +216,7 @@ impl ApiHandler for S3ApiServer {
 				handle_create_multipart_upload(ctx, &req, &key).await
 			}
 			Endpoint::CompleteMultipartUpload { key, upload_id } => {
-				handle_complete_multipart_upload(ctx, req, &key, &upload_id, content_sha256).await
+				handle_complete_multipart_upload(ctx, req, &key, &upload_id).await
 			}
 			Endpoint::CreateBucket {} => unreachable!(),
 			Endpoint::HeadBucket {} => {
@@ -233,6 +226,7 @@ impl ApiHandler for S3ApiServer {
 			Endpoint::DeleteBucket {} => handle_delete_bucket(ctx).await,
 			Endpoint::GetBucketLocation {} => handle_get_bucket_location(ctx),
 			Endpoint::GetBucketVersioning {} => handle_get_bucket_versioning(),
+			Endpoint::GetBucketAcl {} => handle_get_bucket_acl(ctx),
 			Endpoint::ListObjects {
 				delimiter,
 				encoding_type,
@@ -319,7 +313,6 @@ impl ApiHandler for S3ApiServer {
 			} => {
 				let query = ListPartsQuery {
 					bucket_name: ctx.bucket_name.clone(),
-					bucket_id,
 					key,
 					upload_id,
 					part_number_marker: part_number_marker.map(|p| p.min(10000)),
@@ -327,17 +320,15 @@ impl ApiHandler for S3ApiServer {
 				};
 				handle_list_parts(ctx, req, &query).await
 			}
-			Endpoint::DeleteObjects {} => handle_delete_objects(ctx, req, content_sha256).await,
+			Endpoint::DeleteObjects {} => handle_delete_objects(ctx, req).await,
 			Endpoint::GetBucketWebsite {} => handle_get_website(ctx).await,
-			Endpoint::PutBucketWebsite {} => handle_put_website(ctx, req, content_sha256).await,
+			Endpoint::PutBucketWebsite {} => handle_put_website(ctx, req).await,
 			Endpoint::DeleteBucketWebsite {} => handle_delete_website(ctx).await,
 			Endpoint::GetBucketCors {} => handle_get_cors(ctx).await,
-			Endpoint::PutBucketCors {} => handle_put_cors(ctx, req, content_sha256).await,
+			Endpoint::PutBucketCors {} => handle_put_cors(ctx, req).await,
 			Endpoint::DeleteBucketCors {} => handle_delete_cors(ctx).await,
 			Endpoint::GetBucketLifecycleConfiguration {} => handle_get_lifecycle(ctx).await,
-			Endpoint::PutBucketLifecycleConfiguration {} => {
+			Endpoint::PutBucketLifecycleConfiguration {} => handle_put_lifecycle(ctx, req).await,
-				handle_put_lifecycle(ctx, req, content_sha256).await
-			}
 			Endpoint::DeleteBucketLifecycle {} => handle_delete_lifecycle(ctx).await,
 			endpoint => Err(Error::NotImplemented(endpoint.name().to_owned())),
 		};
@@ -352,6 +343,12 @@ impl ApiHandler for S3ApiServer {
 
 		Ok(resp_ok)
 	}
+
+	fn key_id_from_request(&self, req: &Request<IncomingBody>) -> Option<String> {
+		garage_api_common::signature::payload::Authorization::parse_header(req.headers())
+			.map(|auth| auth.key_id)
+			.ok()
+	}
 }
 
 impl ApiEndpoint for S3ApiEndpoint {

src/api/s3/bucket.rs

@@ -1,24 +1,22 @@
 use std::collections::HashMap;
 
-use http_body_util::BodyExt;
 use hyper::{Request, Response, StatusCode};
 
 use garage_model::bucket_alias_table::*;
 use garage_model::bucket_table::Bucket;
 use garage_model::garage::Garage;
-use garage_model::key_table::Key;
+use garage_model::key_table::{Key, KeyParams};
 use garage_model::permission::BucketKeyPerm;
 use garage_table::util::*;
 use garage_util::crdt::*;
-use garage_util::data::*;
 use garage_util::time::*;
 
-use crate::common_error::CommonError;
+use garage_api_common::common_error::CommonError;
-use crate::helpers::*;
+use garage_api_common::helpers::*;
-use crate::s3::api_server::{ReqBody, ResBody};
+
-use crate::s3::error::*;
+use crate::api_server::{ReqBody, ResBody};
-use crate::s3::xml as s3_xml;
+use crate::error::*;
-use crate::signature::verify_signed_content;
+use crate::xml as s3_xml;
 
 pub fn handle_get_bucket_location(ctx: ReqCtx) -> Result<Response<ResBody>, Error> {
 	let ReqCtx { garage, .. } = ctx;
@@ -46,6 +44,55 @@ pub fn handle_get_bucket_versioning() -> Result<Response<ResBody>, Error> {
 		.body(string_body(xml))?)
 }
 
+pub fn handle_get_bucket_acl(ctx: ReqCtx) -> Result<Response<ResBody>, Error> {
+	let ReqCtx {
+		bucket_id, api_key, ..
+	} = ctx;
+	let key_p = api_key.params().ok_or_internal_error(
+		"Key should not be in deleted state at this point (in handle_get_bucket_acl)",
+	)?;
+
+	let mut grants: Vec<s3_xml::Grant> = vec![];
+	let kp = api_key.bucket_permissions(&bucket_id);
+
+	if kp.allow_owner {
+		grants.push(s3_xml::Grant {
+			grantee: create_grantee(&key_p, &api_key),
+			permission: s3_xml::Value("FULL_CONTROL".to_string()),
+		});
+	} else {
+		if kp.allow_read {
+			grants.push(s3_xml::Grant {
+				grantee: create_grantee(&key_p, &api_key),
+				permission: s3_xml::Value("READ".to_string()),
+			});
+			grants.push(s3_xml::Grant {
+				grantee: create_grantee(&key_p, &api_key),
+				permission: s3_xml::Value("READ_ACP".to_string()),
+			});
+		}
+		if kp.allow_write {
+			grants.push(s3_xml::Grant {
+				grantee: create_grantee(&key_p, &api_key),
+				permission: s3_xml::Value("WRITE".to_string()),
+			});
+		}
+	}
+
+	let access_control_policy = s3_xml::AccessControlPolicy {
+		xmlns: (),
+		owner: None,
+		acl: s3_xml::AccessControlList { entries: grants },
+	};
+
+	let xml = s3_xml::to_xml_with_header(&access_control_policy)?;
+	trace!("xml: {}", xml);
+
+	Ok(Response::builder()
+		.header("Content-Type", "application/xml")
+		.body(string_body(xml))?)
+}
+
 pub async fn handle_list_buckets(
 	garage: &Garage,
 	api_key: &Key,
@@ -121,15 +168,10 @@ pub async fn handle_list_buckets(
 pub async fn handle_create_bucket(
 	garage: &Garage,
 	req: Request<ReqBody>,
-	content_sha256: Option<Hash>,
 	api_key_id: &String,
 	bucket_name: String,
 ) -> Result<Response<ResBody>, Error> {
-	let body = BodyExt::collect(req.into_body()).await?.to_bytes();
+	let body = req.into_body().collect().await?;
-
-	if let Some(content_sha256) = content_sha256 {
-		verify_signed_content(content_sha256, &body[..])?;
-	}
 
 	let cmd =
 		parse_create_bucket_xml(&body[..]).ok_or_bad_request("Invalid create bucket XML query")?;
@@ -179,7 +221,7 @@ pub async fn handle_create_bucket(
 		}
 
 		// Create the bucket!
-		if !is_valid_bucket_name(&bucket_name) {
+		if !is_valid_bucket_name(&bucket_name, garage.config.allow_punycode) {
 			return Err(Error::bad_request(format!(
 				"{}: {}",
 				bucket_name, INVALID_BUCKET_NAME_MESSAGE
@@ -248,11 +290,11 @@ pub async fn handle_delete_bucket(ctx: ReqCtx) -> Result<Response<ResBody>, Erro
 		// 1. delete bucket alias
 		if is_local_alias {
 			helper
-				.unset_local_bucket_alias(*bucket_id, &api_key.key_id, bucket_name)
+				.purge_local_bucket_alias(*bucket_id, &api_key.key_id, bucket_name)
 				.await?;
 		} else {
 			helper
-				.unset_global_bucket_alias(*bucket_id, bucket_name)
+				.purge_global_bucket_alias(*bucket_id, bucket_name)
 				.await?;
 		}
 
@@ -318,6 +360,15 @@ fn parse_create_bucket_xml(xml_bytes: &[u8]) -> Option<Option<String>> {
 	Some(ret)
 }
 
+fn create_grantee(key_params: &KeyParams, api_key: &Key) -> s3_xml::Grantee {
+	s3_xml::Grantee {
+		xmlns_xsi: (),
+		typ: "CanonicalUser".to_string(),
+		display_name: Some(s3_xml::Value(key_params.name.get().to_string())),
+		id: Some(s3_xml::Value(api_key.key_id.to_string())),
+	}
+}
+
 #[cfg(test)]
 mod tests {
 	use super::*;

src/api/s3/copy.rs

@@ -1,9 +1,9 @@
 use std::pin::Pin;
-use std::time::{Duration, SystemTime, UNIX_EPOCH};
 
 use futures::{stream, stream::Stream, StreamExt, TryStreamExt};
 
 use bytes::Bytes;
+use http::header::HeaderName;
 use hyper::{Request, Response};
 use serde::Serialize;
 
@@ -20,15 +20,26 @@ use garage_model::s3::mpu_table::*;
 use garage_model::s3::object_table::*;
 use garage_model::s3::version_table::*;
 
-use crate::helpers::*;
+use garage_api_common::helpers::*;
-use crate::s3::api_server::{ReqBody, ResBody};
+use garage_api_common::signature::checksum::*;
-use crate::s3::checksum::*;
+
-use crate::s3::encryption::EncryptionParams;
+use crate::api_server::{ReqBody, ResBody};
-use crate::s3::error::*;
+use crate::encryption::EncryptionParams;
-use crate::s3::get::full_object_byte_stream;
+use crate::error::*;
-use crate::s3::multipart;
+use crate::get::{check_version_not_deleted, full_object_byte_stream, PreconditionHeaders};
-use crate::s3::put::{get_headers, save_stream, ChecksumMode, SaveStreamResult};
+use crate::multipart;
-use crate::s3::xml::{self as s3_xml, xmlns_tag};
+use crate::put::{extract_metadata_headers, save_stream, ChecksumMode, SaveStreamResult};
+use crate::website::X_AMZ_WEBSITE_REDIRECT_LOCATION;
+use crate::xml::{self as s3_xml, xmlns_tag};
+
+pub const X_AMZ_COPY_SOURCE_IF_MATCH: HeaderName =
+	HeaderName::from_static("x-amz-copy-source-if-match");
+pub const X_AMZ_COPY_SOURCE_IF_NONE_MATCH: HeaderName =
+	HeaderName::from_static("x-amz-copy-source-if-none-match");
+pub const X_AMZ_COPY_SOURCE_IF_MODIFIED_SINCE: HeaderName =
+	HeaderName::from_static("x-amz-copy-source-if-modified-since");
+pub const X_AMZ_COPY_SOURCE_IF_UNMODIFIED_SINCE: HeaderName =
+	HeaderName::from_static("x-amz-copy-source-if-unmodified-since");
 
 // -------- CopyObject ---------
 
@@ -37,7 +48,7 @@ pub async fn handle_copy(
 	req: &Request<ReqBody>,
 	dest_key: &str,
 ) -> Result<Response<ResBody>, Error> {
-	let copy_precondition = CopyPreconditionHeaders::parse(req)?;
+	let copy_precondition = PreconditionHeaders::parse_copy_source(req)?;
 
 	let checksum_algorithm = request_checksum_algorithm(req.headers())?;
 
@@ -47,7 +58,7 @@ pub async fn handle_copy(
 		extract_source_info(&source_object)?;
 
 	// Check precondition, e.g. x-amz-copy-source-if-match
-	copy_precondition.check(source_version, &source_version_meta.etag)?;
+	copy_precondition.check_copy_source(source_version, &source_version_meta.etag)?;
 
 	// Determine encryption parameters
 	let (source_encryption, source_object_meta_inner) =
@@ -63,7 +74,7 @@ pub async fn handle_copy(
 	let source_checksum_algorithm = source_checksum.map(|x| x.algorithm());
 
 	// If source object has a checksum, the destination object must as well.
-	// The x-amz-checksum-algorihtm header allows to change that algorithm,
+	// The x-amz-checksum-algorithm header allows to change that algorithm,
 	// but if it is absent, we must use the same as before
 	let checksum_algorithm = checksum_algorithm.or(source_checksum_algorithm);
 
@@ -72,9 +83,20 @@ pub async fn handle_copy(
 	let dest_object_meta = ObjectVersionMetaInner {
 		headers: match req.headers().get("x-amz-metadata-directive") {
 			Some(v) if v == hyper::header::HeaderValue::from_static("REPLACE") => {
-				get_headers(req.headers())?
+				extract_metadata_headers(req.headers())?
+			}
+			_ => {
+				// The x-amz-website-redirect-location header is not copied, instead
+				// it is replaced by the value from the request (or removed if no
+				// value was specified)
+				let is_redirect =
+					|(key, _): &(String, String)| key == X_AMZ_WEBSITE_REDIRECT_LOCATION.as_str();
+				let mut headers: Vec<_> = source_object_meta_inner.headers.clone();
+				headers.retain(|h| !is_redirect(h));
+				let new_headers = extract_metadata_headers(req.headers())?;
+				headers.extend(new_headers.into_iter().filter(is_redirect));
+				headers
 			}
-			_ => source_object_meta_inner.into_owned().headers,
 		},
 		checksum: source_checksum,
 	};
@@ -215,6 +237,7 @@ async fn handle_copy_metaonly(
 				.get(&source_version.uuid, &EmptyKey)
 				.await?;
 			let source_version = source_version.ok_or(Error::NoSuchKey)?;
+			check_version_not_deleted(&source_version)?;
 
 			// Write an "uploading" marker in Object table
 			// This holds a reference to the object in the Version table
@@ -334,7 +357,7 @@ pub async fn handle_upload_part_copy(
 	part_number: u64,
 	upload_id: &str,
 ) -> Result<Response<ResBody>, Error> {
-	let copy_precondition = CopyPreconditionHeaders::parse(req)?;
+	let copy_precondition = PreconditionHeaders::parse_copy_source(req)?;
 
 	let dest_upload_id = multipart::decode_upload_id(upload_id)?;
 
@@ -350,7 +373,7 @@ pub async fn handle_upload_part_copy(
 		extract_source_info(&source_object)?;
 
 	// Check precondition on source, e.g. x-amz-copy-source-if-match
-	copy_precondition.check(source_object_version, &source_version_meta.etag)?;
+	copy_precondition.check_copy_source(source_object_version, &source_version_meta.etag)?;
 
 	// Determine encryption parameters
 	let (source_encryption, _) = EncryptionParams::check_decrypt_for_copy_source(
@@ -406,6 +429,7 @@ pub async fn handle_upload_part_copy(
 		.get(&source_object_version.uuid, &EmptyKey)
 		.await?
 		.ok_or(Error::NoSuchKey)?;
+	check_version_not_deleted(&source_version)?;
 
 	// We want to reuse blocks from the source version as much as possible.
 	// However, we still need to get the data from these blocks
@@ -537,6 +561,7 @@ pub async fn handle_upload_part_copy(
 
 	let mut current_offset = 0;
 	let mut next_block = defragmenter.next().await?;
+	let mut blocks_to_dup = dest_version.clone();
 
 	// TODO this could be optimized similarly to read_and_put_blocks
 	// low priority because uploadpartcopy is rarely used
@@ -566,8 +591,7 @@ pub async fn handle_upload_part_copy(
 			.unwrap()?;
 		checksummer = checksummer_updated;
 
-		dest_version.blocks.clear();
+		let (version_block_key, version_block) = (
-		dest_version.blocks.put(
 			VersionBlockKey {
 				part_number,
 				offset: current_offset,
@@ -579,37 +603,56 @@ pub async fn handle_upload_part_copy(
 		);
 		current_offset += data_len;
 
-		let block_ref = BlockRef {
+		let next = if let Some(final_data) = data_to_upload {
-			block: final_hash,
+			dest_version.blocks.clear();
-			version: dest_version_id,
+			dest_version.blocks.put(version_block_key, version_block);
-			deleted: false.into(),
+			let block_ref = BlockRef {
+				block: final_hash,
+				version: dest_version_id,
+				deleted: false.into(),
+			};
+			let (_, _, _, next) = futures::try_join!(
+				// Thing 1: if the block is not exactly a block that existed before,
+				// we need to insert that data as a new block.
+				garage.block_manager.rpc_put_block(
+					final_hash,
+					final_data,
+					dest_encryption.is_encrypted(),
+					None
+				),
+				// Thing 2: we need to insert the block in the version
+				garage.version_table.insert(&dest_version),
+				// Thing 3: we need to add a block reference
+				garage.block_ref_table.insert(&block_ref),
+				// Thing 4: we need to read the next block
+				defragmenter.next(),
+			)?;
+			next
+		} else {
+			blocks_to_dup.blocks.put(version_block_key, version_block);
+			defragmenter.next().await?
 		};
-
-		let (_, _, _, next) = futures::try_join!(
-			// Thing 1: if the block is not exactly a block that existed before,
-			// we need to insert that data as a new block.
-			async {
-				if let Some(final_data) = data_to_upload {
-					garage
-						.block_manager
-						.rpc_put_block(final_hash, final_data, dest_encryption.is_encrypted(), None)
-						.await
-				} else {
-					Ok(())
-				}
-			},
-			// Thing 2: we need to insert the block in the version
-			garage.version_table.insert(&dest_version),
-			// Thing 3: we need to add a block reference
-			garage.block_ref_table.insert(&block_ref),
-			// Thing 4: we need to read the next block
-			defragmenter.next(),
-		)?;
 		next_block = next;
 	}
 
 	assert_eq!(current_offset, source_range.length);
 
+	// Put the duplicated blocks into the version & block_refs tables
+	let block_refs_to_put = blocks_to_dup
+		.blocks
+		.items()
+		.iter()
+		.map(|b| BlockRef {
+			block: b.1.hash,
+			version: dest_version_id,
+			deleted: false.into(),
+		})
+		.collect::<Vec<_>>();
+	futures::try_join!(
+		garage.version_table.insert(&blocks_to_dup),
+		garage.block_ref_table.insert_many(&block_refs_to_put[..]),
+	)?;
+
 	let checksums = checksummer.finalize();
 	let etag = dest_encryption.etag_from_md5(&checksums.md5);
 	let checksum = checksums.extract(dest_object_checksum_algorithm);
@@ -655,7 +698,8 @@ async fn get_copy_source(ctx: &ReqCtx, req: &Request<ReqBody>) -> Result<Object,
 	let source_bucket_id = garage
 		.bucket_helper()
 		.resolve_bucket(&source_bucket.to_string(), api_key)
-		.await?;
+		.await
+		.map_err(pass_helper_error)?;
 
 	if !api_key.allow_read(&source_bucket_id) {
 		return Err(Error::forbidden(format!(
@@ -701,97 +745,6 @@ fn extract_source_info(
 	Ok((source_version, source_version_data, source_version_meta))
 }
 
-struct CopyPreconditionHeaders {
-	copy_source_if_match: Option<Vec<String>>,
-	copy_source_if_modified_since: Option<SystemTime>,
-	copy_source_if_none_match: Option<Vec<String>>,
-	copy_source_if_unmodified_since: Option<SystemTime>,
-}
-
-impl CopyPreconditionHeaders {
-	fn parse(req: &Request<ReqBody>) -> Result<Self, Error> {
-		Ok(Self {
-			copy_source_if_match: req
-				.headers()
-				.get("x-amz-copy-source-if-match")
-				.map(|x| x.to_str())
-				.transpose()?
-				.map(|x| {
-					x.split(',')
-						.map(|m| m.trim().trim_matches('"').to_string())
-						.collect::<Vec<_>>()
-				}),
-			copy_source_if_modified_since: req
-				.headers()
-				.get("x-amz-copy-source-if-modified-since")
-				.map(|x| x.to_str())
-				.transpose()?
-				.map(httpdate::parse_http_date)
-				.transpose()
-				.ok_or_bad_request("Invalid date in x-amz-copy-source-if-modified-since")?,
-			copy_source_if_none_match: req
-				.headers()
-				.get("x-amz-copy-source-if-none-match")
-				.map(|x| x.to_str())
-				.transpose()?
-				.map(|x| {
-					x.split(',')
-						.map(|m| m.trim().trim_matches('"').to_string())
-						.collect::<Vec<_>>()
-				}),
-			copy_source_if_unmodified_since: req
-				.headers()
-				.get("x-amz-copy-source-if-unmodified-since")
-				.map(|x| x.to_str())
-				.transpose()?
-				.map(httpdate::parse_http_date)
-				.transpose()
-				.ok_or_bad_request("Invalid date in x-amz-copy-source-if-unmodified-since")?,
-		})
-	}
-
-	fn check(&self, v: &ObjectVersion, etag: &str) -> Result<(), Error> {
-		let v_date = UNIX_EPOCH + Duration::from_millis(v.timestamp);
-
-		let ok = match (
-			&self.copy_source_if_match,
-			&self.copy_source_if_unmodified_since,
-			&self.copy_source_if_none_match,
-			&self.copy_source_if_modified_since,
-		) {
-			// TODO I'm not sure all of the conditions are evaluated correctly here
-
-			// If we have both if-match and if-unmodified-since,
-			// basically we don't care about if-unmodified-since,
-			// because in the spec it says that if if-match evaluates to
-			// true but if-unmodified-since evaluates to false,
-			// the copy is still done.
-			(Some(im), _, None, None) => im.iter().any(|x| x == etag || x == "*"),
-			(None, Some(ius), None, None) => v_date <= *ius,
-
-			// If we have both if-none-match and if-modified-since,
-			// then both of the two conditions must evaluate to true
-			(None, None, Some(inm), Some(ims)) => {
-				!inm.iter().any(|x| x == etag || x == "*") && v_date > *ims
-			}
-			(None, None, Some(inm), None) => !inm.iter().any(|x| x == etag || x == "*"),
-			(None, None, None, Some(ims)) => v_date > *ims,
-			(None, None, None, None) => true,
-			_ => {
-				return Err(Error::bad_request(
-					"Invalid combination of x-amz-copy-source-if-xxxxx headers",
-				))
-			}
-		};
-
-		if ok {
-			Ok(())
-		} else {
-			Err(Error::PreconditionFailed)
-		}
-	}
-}
-
 type BlockStreamItemOk = (Bytes, Option<Hash>);
 type BlockStreamItem = Result<BlockStreamItemOk, garage_util::error::Error>;
 
@@ -861,7 +814,7 @@ pub struct CopyPartResult {
 #[cfg(test)]
 mod tests {
 	use super::*;
-	use crate::s3::xml::to_xml_with_header;
+	use crate::xml::to_xml_with_header;
 
 	#[test]
 	fn copy_object_result() -> Result<(), Error> {

src/api/s3/cors.rs

@@ -1,29 +1,16 @@
 use quick_xml::de::from_reader;
-use std::sync::Arc;
 
-use http::header::{
+use hyper::{header::HeaderName, Method, Request, Response, StatusCode};
-	ACCESS_CONTROL_ALLOW_HEADERS, ACCESS_CONTROL_ALLOW_METHODS, ACCESS_CONTROL_ALLOW_ORIGIN,
-	ACCESS_CONTROL_EXPOSE_HEADERS, ACCESS_CONTROL_REQUEST_HEADERS, ACCESS_CONTROL_REQUEST_METHOD,
-};
-use hyper::{
-	body::Body, body::Incoming as IncomingBody, header::HeaderName, Method, Request, Response,
-	StatusCode,
-};
-
-use http_body_util::BodyExt;
 
 use serde::{Deserialize, Serialize};
 
-use crate::common_error::CommonError;
+use garage_model::bucket_table::{Bucket, CorsRule as GarageCorsRule};
-use crate::helpers::*;
-use crate::s3::api_server::{ReqBody, ResBody};
-use crate::s3::error::*;
-use crate::s3::xml::{to_xml_with_header, xmlns_tag, IntValue, Value};
-use crate::signature::verify_signed_content;
 
-use garage_model::bucket_table::{Bucket, BucketParams, CorsRule as GarageCorsRule};
+use garage_api_common::helpers::*;
-use garage_model::garage::Garage;
+
-use garage_util::data::*;
+use crate::api_server::{ReqBody, ResBody};
+use crate::error::*;
+use crate::xml::{to_xml_with_header, xmlns_tag, IntValue, Value};
 
 pub async fn handle_get_cors(ctx: ReqCtx) -> Result<Response<ResBody>, Error> {
 	let ReqCtx { bucket_params, .. } = ctx;
@@ -68,7 +55,6 @@ pub async fn handle_delete_cors(ctx: ReqCtx) -> Result<Response<ResBody>, Error>
 pub async fn handle_put_cors(
 	ctx: ReqCtx,
 	req: Request<ReqBody>,
-	content_sha256: Option<Hash>,
 ) -> Result<Response<ResBody>, Error> {
 	let ReqCtx {
 		garage,
@@ -77,11 +63,7 @@ pub async fn handle_put_cors(
 		..
 	} = ctx;
 
-	let body = BodyExt::collect(req.into_body()).await?.to_bytes();
+	let body = req.into_body().collect().await?;
-
-	if let Some(content_sha256) = content_sha256 {
-		verify_signed_content(content_sha256, &body[..])?;
-	}
 
 	let conf: CorsConfiguration = from_reader(&body as &[u8])?;
 	conf.validate()?;
@@ -99,154 +81,6 @@ pub async fn handle_put_cors(
 		.body(empty_body())?)
 }
 
-pub async fn handle_options_api(
-	garage: Arc<Garage>,
-	req: &Request<IncomingBody>,
-	bucket_name: Option<String>,
-) -> Result<Response<EmptyBody>, CommonError> {
-	// FIXME: CORS rules of buckets with local aliases are
-	// not taken into account.
-
-	// If the bucket name is a global bucket name,
-	// we try to apply the CORS rules of that bucket.
-	// If a user has a local bucket name that has
-	// the same name, its CORS rules won't be applied
-	// and will be shadowed by the rules of the globally
-	// existing bucket (but this is inevitable because
-	// OPTIONS calls are not auhtenticated).
-	if let Some(bn) = bucket_name {
-		let helper = garage.bucket_helper();
-		let bucket_id = helper.resolve_global_bucket_name(&bn).await?;
-		if let Some(id) = bucket_id {
-			let bucket = garage.bucket_helper().get_existing_bucket(id).await?;
-			let bucket_params = bucket.state.into_option().unwrap();
-			handle_options_for_bucket(req, &bucket_params)
-		} else {
-			// If there is a bucket name in the request, but that name
-			// does not correspond to a global alias for a bucket,
-			// then it's either a non-existing bucket or a local bucket.
-			// We have no way of knowing, because the request is not
-			// authenticated and thus we can't resolve local aliases.
-			// We take the permissive approach of allowing everything,
-			// because we don't want to prevent web apps that use
-			// local bucket names from making API calls.
-			Ok(Response::builder()
-				.header(ACCESS_CONTROL_ALLOW_ORIGIN, "*")
-				.header(ACCESS_CONTROL_ALLOW_METHODS, "*")
-				.status(StatusCode::OK)
-				.body(EmptyBody::new())?)
-		}
-	} else {
-		// If there is no bucket name in the request,
-		// we are doing a ListBuckets call, which we want to allow
-		// for all origins.
-		Ok(Response::builder()
-			.header(ACCESS_CONTROL_ALLOW_ORIGIN, "*")
-			.header(ACCESS_CONTROL_ALLOW_METHODS, "GET")
-			.status(StatusCode::OK)
-			.body(EmptyBody::new())?)
-	}
-}
-
-pub fn handle_options_for_bucket(
-	req: &Request<IncomingBody>,
-	bucket_params: &BucketParams,
-) -> Result<Response<EmptyBody>, CommonError> {
-	let origin = req
-		.headers()
-		.get("Origin")
-		.ok_or_bad_request("Missing Origin header")?
-		.to_str()?;
-	let request_method = req
-		.headers()
-		.get(ACCESS_CONTROL_REQUEST_METHOD)
-		.ok_or_bad_request("Missing Access-Control-Request-Method header")?
-		.to_str()?;
-	let request_headers = match req.headers().get(ACCESS_CONTROL_REQUEST_HEADERS) {
-		Some(h) => h.to_str()?.split(',').map(|h| h.trim()).collect::<Vec<_>>(),
-		None => vec![],
-	};
-
-	if let Some(cors_config) = bucket_params.cors_config.get() {
-		let matching_rule = cors_config
-			.iter()
-			.find(|rule| cors_rule_matches(rule, origin, request_method, request_headers.iter()));
-		if let Some(rule) = matching_rule {
-			let mut resp = Response::builder()
-				.status(StatusCode::OK)
-				.body(EmptyBody::new())?;
-			add_cors_headers(&mut resp, rule).ok_or_internal_error("Invalid CORS configuration")?;
-			return Ok(resp);
-		}
-	}
-
-	Err(CommonError::Forbidden(
-		"This CORS request is not allowed.".into(),
-	))
-}
-
-pub fn find_matching_cors_rule<'a>(
-	bucket_params: &'a BucketParams,
-	req: &Request<impl Body>,
-) -> Result<Option<&'a GarageCorsRule>, Error> {
-	if let Some(cors_config) = bucket_params.cors_config.get() {
-		if let Some(origin) = req.headers().get("Origin") {
-			let origin = origin.to_str()?;
-			let request_headers = match req.headers().get(ACCESS_CONTROL_REQUEST_HEADERS) {
-				Some(h) => h.to_str()?.split(',').map(|h| h.trim()).collect::<Vec<_>>(),
-				None => vec![],
-			};
-			return Ok(cors_config.iter().find(|rule| {
-				cors_rule_matches(rule, origin, req.method().as_ref(), request_headers.iter())
-			}));
-		}
-	}
-	Ok(None)
-}
-
-fn cors_rule_matches<'a, HI, S>(
-	rule: &GarageCorsRule,
-	origin: &'a str,
-	method: &'a str,
-	mut request_headers: HI,
-) -> bool
-where
-	HI: Iterator<Item = S>,
-	S: AsRef<str>,
-{
-	rule.allow_origins.iter().any(|x| x == "*" || x == origin)
-		&& rule.allow_methods.iter().any(|x| x == "*" || x == method)
-		&& request_headers.all(|h| {
-			rule.allow_headers
-				.iter()
-				.any(|x| x == "*" || x == h.as_ref())
-		})
-}
-
-pub fn add_cors_headers(
-	resp: &mut Response<impl Body>,
-	rule: &GarageCorsRule,
-) -> Result<(), http::header::InvalidHeaderValue> {
-	let h = resp.headers_mut();
-	h.insert(
-		ACCESS_CONTROL_ALLOW_ORIGIN,
-		rule.allow_origins.join(", ").parse()?,
-	);
-	h.insert(
-		ACCESS_CONTROL_ALLOW_METHODS,
-		rule.allow_methods.join(", ").parse()?,
-	);
-	h.insert(
-		ACCESS_CONTROL_ALLOW_HEADERS,
-		rule.allow_headers.join(", ").parse()?,
-	);
-	h.insert(
-		ACCESS_CONTROL_EXPOSE_HEADERS,
-		rule.expose_headers.join(", ").parse()?,
-	);
-	Ok(())
-}
-
 // ---- SERIALIZATION AND DESERIALIZATION TO/FROM S3 XML ----
 
 #[derive(Debug, Serialize, Deserialize, PartialEq, Eq, PartialOrd, Ord)]
@@ -254,7 +88,9 @@ pub fn add_cors_headers(
 pub struct CorsConfiguration {
 	#[serde(serialize_with = "xmlns_tag", skip_deserializing)]
 	pub xmlns: (),
-	#[serde(rename = "CORSRule")]
+	// "default" is required to be able to parse an empty list of rules,
+	// cf https://docs.rs/quick-xml/latest/quick_xml/de/#sequences-xsall-and-xssequence-xml-schema-types
+	#[serde(rename = "CORSRule", default)]
 	pub cors_rules: Vec<CorsRule>,
 }
 
@@ -436,4 +272,26 @@ mod tests {
 
 		Ok(())
 	}
+
+	#[test]
+	fn test_deserialize_norules() -> Result<(), Error> {
+		let message = r#"<?xml version="1.0" encoding="UTF-8"?>
+<CORSConfiguration xmlns="http://s3.amazonaws.com/doc/2006-03-01/" />"#;
+		let conf: CorsConfiguration = from_str(message).unwrap();
+		let ref_value = CorsConfiguration {
+			xmlns: (),
+			cors_rules: vec![],
+		};
+		assert_eq! {
+			ref_value,
+			conf
+		};
+
+		let message2 = to_xml_with_header(&ref_value)?;
+
+		let cleanup = |c: &str| c.replace(char::is_whitespace, "");
+		assert_eq!(cleanup(message), cleanup(&message2));
+
+		Ok(())
+	}
 }

src/api/s3/delete.rs

@@ -1,16 +1,15 @@
-use http_body_util::BodyExt;
 use hyper::{Request, Response, StatusCode};
 
 use garage_util::data::*;
 
 use garage_model::s3::object_table::*;
 
-use crate::helpers::*;
+use garage_api_common::helpers::*;
-use crate::s3::api_server::{ReqBody, ResBody};
+
-use crate::s3::error::*;
+use crate::api_server::{ReqBody, ResBody};
-use crate::s3::put::next_timestamp;
+use crate::error::*;
-use crate::s3::xml as s3_xml;
+use crate::put::next_timestamp;
-use crate::signature::verify_signed_content;
+use crate::xml as s3_xml;
 
 async fn handle_delete_internal(ctx: &ReqCtx, key: &str) -> Result<(Uuid, Uuid), Error> {
 	let ReqCtx {
@@ -67,13 +66,8 @@ pub async fn handle_delete(ctx: ReqCtx, key: &str) -> Result<Response<ResBody>,
 pub async fn handle_delete_objects(
 	ctx: ReqCtx,
 	req: Request<ReqBody>,
-	content_sha256: Option<Hash>,
 ) -> Result<Response<ResBody>, Error> {
-	let body = BodyExt::collect(req.into_body()).await?.to_bytes();
+	let body = req.into_body().collect().await?;
-
-	if let Some(content_sha256) = content_sha256 {
-		verify_signed_content(content_sha256, &body[..])?;
-	}
 
 	let cmd_xml = roxmltree::Document::parse(std::str::from_utf8(&body)?)?;
 	let cmd = parse_delete_objects_xml(&cmd_xml).ok_or_bad_request("Invalid delete XML query")?;

src/api/s3/encryption.rs

@@ -28,9 +28,10 @@ use garage_util::migrate::Migrate;
 use garage_model::garage::Garage;
 use garage_model::s3::object_table::{ObjectVersionEncryption, ObjectVersionMetaInner};
 
-use crate::common_error::*;
+use garage_api_common::common_error::*;
-use crate::s3::checksum::Md5Checksum;
+use garage_api_common::signature::checksum::Md5Checksum;
-use crate::s3::error::Error;
+
+use crate::error::Error;
 
 const X_AMZ_SERVER_SIDE_ENCRYPTION_CUSTOMER_ALGORITHM: HeaderName =
 	HeaderName::from_static("x-amz-server-side-encryption-customer-algorithm");

src/api/s3/error.rs

@@ -1,93 +1,109 @@
 use std::convert::TryInto;
 
-use err_derive::Error;
 use hyper::header::HeaderValue;
 use hyper::{HeaderMap, StatusCode};
+use thiserror::Error;
 
-use crate::common_error::CommonError;
+use garage_model::helper::error::Error as HelperError;
-pub use crate::common_error::{CommonErrorDerivative, OkOrBadRequest, OkOrInternalError};
+
-use crate::generic_server::ApiError;
+pub(crate) use garage_api_common::common_error::pass_helper_error;
-use crate::helpers::*;
+
-use crate::s3::xml as s3_xml;
+use garage_api_common::common_error::{
-use crate::signature::error::Error as SignatureError;
+	commonErrorDerivative, helper_error_as_internal, CommonError,
+};
+
+pub use garage_api_common::common_error::{
+	CommonErrorDerivative, OkOrBadRequest, OkOrInternalError,
+};
+
+use garage_api_common::generic_server::ApiError;
+use garage_api_common::helpers::*;
+use garage_api_common::signature::error::Error as SignatureError;
+
+use crate::xml as s3_xml;
 
 /// Errors of this crate
 #[derive(Debug, Error)]
 pub enum Error {
-	#[error(display = "{}", _0)]
+	#[error("{0}")]
 	/// Error from common error
-	Common(CommonError),
+	Common(#[from] CommonError),
 
 	// Category: cannot process
 	/// Authorization Header Malformed
-	#[error(display = "Authorization header malformed, unexpected scope: {}", _0)]
+	#[error("Authorization header malformed, unexpected scope: {0}")]
 	AuthorizationHeaderMalformed(String),
 
 	/// The object requested don't exists
-	#[error(display = "Key not found")]
+	#[error("Key not found")]
 	NoSuchKey,
 
 	/// The multipart upload requested don't exists
-	#[error(display = "Upload not found")]
+	#[error("Upload not found")]
 	NoSuchUpload,
 
 	/// Precondition failed (e.g. x-amz-copy-source-if-match)
-	#[error(display = "At least one of the preconditions you specified did not hold")]
+	#[error("At least one of the preconditions you specified did not hold")]
 	PreconditionFailed,
 
 	/// Parts specified in CMU request do not match parts actually uploaded
-	#[error(display = "Parts given to CompleteMultipartUpload do not match uploaded parts")]
+	#[error("Parts given to CompleteMultipartUpload do not match uploaded parts")]
 	InvalidPart,
 
 	/// Parts given to CompleteMultipartUpload were not in ascending order
-	#[error(display = "Parts given to CompleteMultipartUpload were not in ascending order")]
+	#[error("Parts given to CompleteMultipartUpload were not in ascending order")]
 	InvalidPartOrder,
 
 	/// In CompleteMultipartUpload: not enough data
 	/// (here we are more lenient than AWS S3)
-	#[error(display = "Proposed upload is smaller than the minimum allowed object size")]
+	#[error("Proposed upload is smaller than the minimum allowed object size")]
 	EntityTooSmall,
 
 	// Category: bad request
 	/// The request contained an invalid UTF-8 sequence in its path or in other parameters
-	#[error(display = "Invalid UTF-8: {}", _0)]
+	#[error("Invalid UTF-8: {0}")]
-	InvalidUtf8Str(#[error(source)] std::str::Utf8Error),
+	InvalidUtf8Str(#[from] std::str::Utf8Error),
 
 	/// The request used an invalid path
-	#[error(display = "Invalid UTF-8: {}", _0)]
+	#[error("Invalid UTF-8: {0}")]
-	InvalidUtf8String(#[error(source)] std::string::FromUtf8Error),
+	InvalidUtf8String(#[from] std::string::FromUtf8Error),
 
 	/// The client sent invalid XML data
-	#[error(display = "Invalid XML: {}", _0)]
+	#[error("Invalid XML: {0}")]
 	InvalidXml(String),
 
 	/// The client sent a range header with invalid value
-	#[error(display = "Invalid HTTP range: {:?}", _0)]
+	#[error("Invalid HTTP range: {0:?}")]
-	InvalidRange(#[error(from)] (http_range::HttpRangeParseError, u64)),
+	InvalidRange((http_range::HttpRangeParseError, u64)),
 
 	/// The client sent a range header with invalid value
-	#[error(display = "Invalid encryption algorithm: {:?}, should be AES256", _0)]
+	#[error("Invalid encryption algorithm: {0:?}, should be AES256")]
 	InvalidEncryptionAlgorithm(String),
 
-	/// The client sent invalid XML data
+	/// The provided digest (checksum) value was invalid
-	#[error(display = "Invalid digest: {}", _0)]
+	#[error("Invalid digest: {0}")]
 	InvalidDigest(String),
 
 	/// The client sent a request for an action not supported by garage
-	#[error(display = "Unimplemented action: {}", _0)]
+	#[error("Unimplemented action: {0}")]
 	NotImplemented(String),
 }
 
-impl<T> From<T> for Error
+commonErrorDerivative!(Error);
-where
+
-	CommonError: From<T>,
+// Helper errors are always passed as internal errors by default.
-{
+// To pass the specific error code back to the client, use `pass_helper_error`.
-	fn from(err: T) -> Self {
+impl From<HelperError> for Error {
-		Error::Common(CommonError::from(err))
+	fn from(err: HelperError) -> Error {
+		Error::Common(helper_error_as_internal(err))
 	}
 }
 
-impl CommonErrorDerivative for Error {}
+impl From<(http_range::HttpRangeParseError, u64)> for Error {
+	fn from(err: (http_range::HttpRangeParseError, u64)) -> Error {
+		Error::InvalidRange(err)
+	}
+}
 
 impl From<roxmltree::Error> for Error {
 	fn from(err: roxmltree::Error) -> Self {
@@ -109,6 +125,7 @@ impl From<SignatureError> for Error {
 				Self::AuthorizationHeaderMalformed(c)
 			}
 			SignatureError::InvalidUtf8Str(i) => Self::InvalidUtf8Str(i),
+			SignatureError::InvalidDigest(d) => Self::InvalidDigest(d),
 		}
 	}
 }

src/api/s3/get.rs

@@ -2,36 +2,39 @@
 use std::collections::BTreeMap;
 use std::convert::TryInto;
 use std::sync::Arc;
-use std::time::{Duration, UNIX_EPOCH};
+use std::time::{Duration, SystemTime, UNIX_EPOCH};
 
 use bytes::Bytes;
 use futures::future;
 use futures::stream::{self, Stream, StreamExt};
 use http::header::{
-	ACCEPT_RANGES, CACHE_CONTROL, CONTENT_DISPOSITION, CONTENT_ENCODING, CONTENT_LANGUAGE,
+	HeaderMap, HeaderName, ACCEPT_RANGES, CACHE_CONTROL, CONTENT_DISPOSITION, CONTENT_ENCODING,
-	CONTENT_LENGTH, CONTENT_RANGE, CONTENT_TYPE, ETAG, EXPIRES, IF_MODIFIED_SINCE, IF_NONE_MATCH,
+	CONTENT_LANGUAGE, CONTENT_LENGTH, CONTENT_RANGE, CONTENT_TYPE, ETAG, EXPIRES, IF_MATCH,
-	LAST_MODIFIED, RANGE,
+	IF_MODIFIED_SINCE, IF_NONE_MATCH, IF_UNMODIFIED_SINCE, LAST_MODIFIED, RANGE,
 };
-use hyper::{body::Body, Request, Response, StatusCode};
+use hyper::{Request, Response, StatusCode};
 use tokio::sync::mpsc;
 
 use garage_net::stream::ByteStream;
 use garage_rpc::rpc_helper::OrderTag;
 use garage_table::EmptyKey;
 use garage_util::data::*;
-use garage_util::error::OkOrMessage;
+use garage_util::error::{Error as UtilError, OkOrMessage};
 
 use garage_model::garage::Garage;
 use garage_model::s3::object_table::*;
 use garage_model::s3::version_table::*;
 
-use crate::helpers::*;
+use garage_api_common::common_error::CommonError;
-use crate::s3::api_server::ResBody;
+use garage_api_common::helpers::*;
-use crate::s3::checksum::{add_checksum_response_headers, X_AMZ_CHECKSUM_MODE};
+use garage_api_common::signature::checksum::{add_checksum_response_headers, X_AMZ_CHECKSUM_MODE};
-use crate::s3::encryption::EncryptionParams;
-use crate::s3::error::*;
 
-const X_AMZ_MP_PARTS_COUNT: &str = "x-amz-mp-parts-count";
+use crate::api_server::ResBody;
+use crate::copy::*;
+use crate::encryption::EncryptionParams;
+use crate::error::*;
+
+const X_AMZ_MP_PARTS_COUNT: HeaderName = HeaderName::from_static("x-amz-mp-parts-count");
 
 #[derive(Default)]
 pub struct GetObjectOverrides {
@@ -68,14 +71,11 @@ fn object_headers(
 	// See: https://docs.aws.amazon.com/AmazonS3/latest/userguide/UsingMetadata.html
 	let mut headers_by_name = BTreeMap::new();
 	for (name, value) in meta_inner.headers.iter() {
-		match headers_by_name.get_mut(name) {
+		let name_lower = name.to_ascii_lowercase();
-			None => {
+		headers_by_name
-				headers_by_name.insert(name, vec![value.as_str()]);
+			.entry(name_lower)
-			}
+			.or_insert(vec![])
-			Some(headers) => {
+			.push(value.as_str());
-				headers.push(value.as_str());
-			}
-		}
 	}
 
 	for (name, values) in headers_by_name {
@@ -117,49 +117,29 @@ fn getobject_override_headers(
 	Ok(())
 }
 
-fn try_answer_cached(
+fn handle_http_precondition(
 	version: &ObjectVersion,
 	version_meta: &ObjectVersionMeta,
-	req: &Request<impl Body>,
+	req: &Request<()>,
-) -> Option<Response<ResBody>> {
+) -> Result<Option<Response<ResBody>>, Error> {
-	// <trinity> It is possible, and is even usually the case, [that both If-None-Match and
+	let precondition_headers = PreconditionHeaders::parse(req)?;
-	// If-Modified-Since] are present in a request. In this situation If-None-Match takes
-	// precedence and If-Modified-Since is ignored (as per 6.Precedence from rfc7232). The rational
-	// being that etag based matching is more accurate, it has no issue with sub-second precision
-	// for instance (in case of very fast updates)
-	let cached = if let Some(none_match) = req.headers().get(IF_NONE_MATCH) {
-		let none_match = none_match.to_str().ok()?;
-		let expected = format!("\"{}\"", version_meta.etag);
-		let found = none_match
-			.split(',')
-			.map(str::trim)
-			.any(|etag| etag == expected || etag == "\"*\"");
-		found
-	} else if let Some(modified_since) = req.headers().get(IF_MODIFIED_SINCE) {
-		let modified_since = modified_since.to_str().ok()?;
-		let client_date = httpdate::parse_http_date(modified_since).ok()?;
-		let server_date = UNIX_EPOCH + Duration::from_millis(version.timestamp);
-		client_date >= server_date
-	} else {
-		false
-	};
 
-	if cached {
+	if let Some(status_code) = precondition_headers.check(&version, &version_meta.etag)? {
-		Some(
+		Ok(Some(
 			Response::builder()
-				.status(StatusCode::NOT_MODIFIED)
+				.status(status_code)
 				.body(empty_body())
 				.unwrap(),
-		)
+		))
 	} else {
-		None
+		Ok(None)
 	}
 }
 
 /// Handle HEAD request
 pub async fn handle_head(
 	ctx: ReqCtx,
-	req: &Request<impl Body>,
+	req: &Request<()>,
 	key: &str,
 	part_number: Option<u64>,
 ) -> Result<Response<ResBody>, Error> {
@@ -169,7 +149,7 @@ pub async fn handle_head(
 /// Handle HEAD request for website
 pub async fn handle_head_without_ctx(
 	garage: Arc<Garage>,
-	req: &Request<impl Body>,
+	req: &Request<()>,
 	bucket_id: Uuid,
 	key: &str,
 	part_number: Option<u64>,
@@ -198,8 +178,8 @@ pub async fn handle_head_without_ctx(
 		_ => unreachable!(),
 	};
 
-	if let Some(cached) = try_answer_cached(object_version, version_meta, req) {
+	if let Some(res) = handle_http_precondition(object_version, version_meta, req)? {
-		return Ok(cached);
+		return Ok(res);
 	}
 
 	let (encryption, headers) =
@@ -236,6 +216,7 @@ pub async fn handle_head_without_ctx(
 					.get(&object_version.uuid, &EmptyKey)
 					.await?
 					.ok_or(Error::NoSuchKey)?;
+				check_version_not_deleted(&version)?;
 
 				let (part_offset, part_end) =
 					calculate_part_bounds(&version, pn).ok_or(Error::InvalidPart)?;
@@ -280,7 +261,7 @@ pub async fn handle_head_without_ctx(
 /// Handle GET request
 pub async fn handle_get(
 	ctx: ReqCtx,
-	req: &Request<impl Body>,
+	req: &Request<()>,
 	key: &str,
 	part_number: Option<u64>,
 	overrides: GetObjectOverrides,
@@ -291,7 +272,7 @@ pub async fn handle_get(
 /// Handle GET request
 pub async fn handle_get_without_ctx(
 	garage: Arc<Garage>,
-	req: &Request<impl Body>,
+	req: &Request<()>,
 	bucket_id: Uuid,
 	key: &str,
 	part_number: Option<u64>,
@@ -320,8 +301,8 @@ pub async fn handle_get_without_ctx(
 		ObjectVersionData::FirstBlock(meta, _) => meta,
 	};
 
-	if let Some(cached) = try_answer_cached(last_v, last_v_meta, req) {
+	if let Some(res) = handle_http_precondition(last_v, last_v_meta, req)? {
-		return Ok(cached);
+		return Ok(res);
 	}
 
 	let (enc, headers) =
@@ -342,7 +323,12 @@ pub async fn handle_get_without_ctx(
 				enc,
 				&headers,
 				pn,
-				checksum_mode,
+				ChecksumMode {
+					// TODO: for multipart uploads, checksums of each part should be stored
+					// so that we can return the corresponding checksum here
+					// https://docs.aws.amazon.com/AmazonS3/latest/userguide/checking-object-integrity.html
+					enabled: false,
+				},
 			)
 			.await
 		}
@@ -356,7 +342,12 @@ pub async fn handle_get_without_ctx(
 				&headers,
 				range.start,
 				range.start + range.length,
-				checksum_mode,
+				ChecksumMode {
+					// TODO: for range queries that align with part boundaries,
+					// we should return the saved checksum of the part
+					// https://docs.aws.amazon.com/AmazonS3/latest/userguide/checking-object-integrity.html
+					enabled: false,
+				},
 			)
 			.await
 		}
@@ -376,6 +367,21 @@ pub async fn handle_get_without_ctx(
 	}
 }
 
+pub(crate) fn check_version_not_deleted(version: &Version) -> Result<(), Error> {
+	if version.deleted.get() {
+		// the version was deleted between when the object_table was consulted
+		// and now, this could mean the object was deleted, or overriden.
+		// Rather than say the key doesn't exist, return a transient error
+		// to signal the client to try again.
+		return Err(CommonError::InternalError(UtilError::Message(
+			"conflict/inconsistency between object and version state, version is deleted"
+				.to_string(),
+		))
+		.into());
+	}
+	Ok(())
+}
+
 async fn handle_get_full(
 	garage: Arc<Garage>,
 	version: &ObjectVersion,
@@ -442,6 +448,7 @@ pub fn full_object_byte_stream(
 						.ok_or_message("channel closed")?;
 
 					let version = version_fut.await.unwrap()?.ok_or(Error::NoSuchKey)?;
+					check_version_not_deleted(&version)?;
 					for (i, (_, vb)) in version.blocks.items().iter().enumerate().skip(1) {
 						let stream_block_i = encryption
 							.get_block(&garage, &vb.hash, Some(order_stream.order(i as u64)))
@@ -457,6 +464,14 @@ pub fn full_object_byte_stream(
 				{
 					Ok(()) => (),
 					Err(e) => {
+						// TODO i think this is a bad idea, we should log
+						// an error and stop there. If the error happens to
+						// be exactly the size of what hasn't been streamed
+						// yet, the client will see the request as a
+						// success
+						// instead truncating the output notify the client
+						// something happened with their download, so that
+						// they can retry it
 						let _ = tx.send(error_stream_item(e)).await;
 					}
 				}
@@ -508,7 +523,7 @@ async fn handle_get_range(
 				.get(&version.uuid, &EmptyKey)
 				.await?
 				.ok_or(Error::NoSuchKey)?;
-
+			check_version_not_deleted(&version)?;
 			let body =
 				body_from_blocks_range(garage, encryption, version.blocks.items(), begin, end);
 			Ok(resp_builder.body(body)?)
@@ -559,6 +574,8 @@ async fn handle_get_part(
 				.await?
 				.ok_or(Error::NoSuchKey)?;
 
+			check_version_not_deleted(&version)?;
+
 			let (begin, end) =
 				calculate_part_bounds(&version, part_number).ok_or(Error::InvalidPart)?;
 
@@ -579,7 +596,7 @@ async fn handle_get_part(
 }
 
 fn parse_range_header(
-	req: &Request<impl Body>,
+	req: &Request<()>,
 	total_size: u64,
 ) -> Result<Option<http_range::HttpRange>, Error> {
 	let range = match req.headers().get(RANGE) {
@@ -620,7 +637,7 @@ struct ChecksumMode {
 	enabled: bool,
 }
 
-fn checksum_mode(req: &Request<impl Body>) -> ChecksumMode {
+fn checksum_mode(req: &Request<()>) -> ChecksumMode {
 	ChecksumMode {
 		enabled: req
 			.headers()
@@ -753,3 +770,118 @@ fn std_error_from_read_error<E: std::fmt::Display>(e: E) -> std::io::Error {
 		format!("Error while reading object data: {}", e),
 	)
 }
+
+// ----
+
+pub struct PreconditionHeaders {
+	if_match: Option<Vec<String>>,
+	if_modified_since: Option<SystemTime>,
+	if_none_match: Option<Vec<String>>,
+	if_unmodified_since: Option<SystemTime>,
+}
+
+impl PreconditionHeaders {
+	fn parse<B>(req: &Request<B>) -> Result<Self, Error> {
+		Self::parse_with(
+			req.headers(),
+			&IF_MATCH,
+			&IF_NONE_MATCH,
+			&IF_MODIFIED_SINCE,
+			&IF_UNMODIFIED_SINCE,
+		)
+	}
+
+	pub(crate) fn parse_copy_source<B>(req: &Request<B>) -> Result<Self, Error> {
+		Self::parse_with(
+			req.headers(),
+			&X_AMZ_COPY_SOURCE_IF_MATCH,
+			&X_AMZ_COPY_SOURCE_IF_NONE_MATCH,
+			&X_AMZ_COPY_SOURCE_IF_MODIFIED_SINCE,
+			&X_AMZ_COPY_SOURCE_IF_UNMODIFIED_SINCE,
+		)
+	}
+
+	fn parse_with(
+		headers: &HeaderMap,
+		hdr_if_match: &HeaderName,
+		hdr_if_none_match: &HeaderName,
+		hdr_if_modified_since: &HeaderName,
+		hdr_if_unmodified_since: &HeaderName,
+	) -> Result<Self, Error> {
+		Ok(Self {
+			if_match: headers
+				.get(hdr_if_match)
+				.map(|x| x.to_str())
+				.transpose()?
+				.map(|x| {
+					x.split(',')
+						.map(|m| m.trim().trim_matches('"').to_string())
+						.collect::<Vec<_>>()
+				}),
+			if_none_match: headers
+				.get(hdr_if_none_match)
+				.map(|x| x.to_str())
+				.transpose()?
+				.map(|x| {
+					x.split(',')
+						.map(|m| m.trim().trim_matches('"').to_string())
+						.collect::<Vec<_>>()
+				}),
+			if_modified_since: headers
+				.get(hdr_if_modified_since)
+				.map(|x| x.to_str())
+				.transpose()?
+				.map(httpdate::parse_http_date)
+				.transpose()
+				.ok_or_bad_request("Invalid date in if-modified-since")?,
+			if_unmodified_since: headers
+				.get(hdr_if_unmodified_since)
+				.map(|x| x.to_str())
+				.transpose()?
+				.map(httpdate::parse_http_date)
+				.transpose()
+				.ok_or_bad_request("Invalid date in if-unmodified-since")?,
+		})
+	}
+
+	fn check(&self, v: &ObjectVersion, etag: &str) -> Result<Option<StatusCode>, Error> {
+		// we store date with ms precision, but headers are precise to the second: truncate
+		// the timestamp to handle the same-second edge case
+		let v_date = UNIX_EPOCH + Duration::from_secs(v.timestamp / 1000);
+
+		// Implemented from https://datatracker.ietf.org/doc/html/rfc7232#section-6
+
+		if let Some(im) = &self.if_match {
+			// Step 1: if-match is present
+			if !im.iter().any(|x| x == etag || x == "*") {
+				return Ok(Some(StatusCode::PRECONDITION_FAILED));
+			}
+		} else if let Some(ius) = &self.if_unmodified_since {
+			// Step 2: if-unmodified-since is present, and if-match is absent
+			if v_date > *ius {
+				return Ok(Some(StatusCode::PRECONDITION_FAILED));
+			}
+		}
+
+		if let Some(inm) = &self.if_none_match {
+			// Step 3: if-none-match is present
+			if inm.iter().any(|x| x == etag || x == "*") {
+				return Ok(Some(StatusCode::NOT_MODIFIED));
+			}
+		} else if let Some(ims) = &self.if_modified_since {
+			// Step 4: if-modified-since is present, and if-none-match is absent
+			if v_date <= *ims {
+				return Ok(Some(StatusCode::NOT_MODIFIED));
+			}
+		}
+
+		Ok(None)
+	}
+
+	pub(crate) fn check_copy_source(&self, v: &ObjectVersion, etag: &str) -> Result<(), Error> {
+		match self.check(v, etag)? {
+			Some(_) => Err(Error::PreconditionFailed),
+			None => Ok(()),
+		}
+	}
+}

src/api/s3/lib.rs

@@ -1,3 +1,6 @@
+#[macro_use]
+extern crate tracing;
+
 pub mod api_server;
 pub mod error;
 
@@ -11,9 +14,8 @@ mod list;
 mod multipart;
 mod post_object;
 mod put;
-mod website;
+pub mod website;
 
-mod checksum;
 mod encryption;
 mod router;
 pub mod xml;

src/api/s3/lifecycle.rs

@@ -1,21 +1,19 @@
 use quick_xml::de::from_reader;
 
-use http_body_util::BodyExt;
 use hyper::{Request, Response, StatusCode};
 
 use serde::{Deserialize, Serialize};
 
-use crate::helpers::*;
+use garage_api_common::helpers::*;
-use crate::s3::api_server::{ReqBody, ResBody};
+
-use crate::s3::error::*;
+use crate::api_server::{ReqBody, ResBody};
-use crate::s3::xml::{to_xml_with_header, xmlns_tag, IntValue, Value};
+use crate::error::*;
-use crate::signature::verify_signed_content;
+use crate::xml::{to_xml_with_header, xmlns_tag, IntValue, Value};
 
 use garage_model::bucket_table::{
 	parse_lifecycle_date, Bucket, LifecycleExpiration as GarageLifecycleExpiration,
 	LifecycleFilter as GarageLifecycleFilter, LifecycleRule as GarageLifecycleRule,
 };
-use garage_util::data::*;
 
 pub async fn handle_get_lifecycle(ctx: ReqCtx) -> Result<Response<ResBody>, Error> {
 	let ReqCtx { bucket_params, .. } = ctx;
@@ -29,7 +27,7 @@ pub async fn handle_get_lifecycle(ctx: ReqCtx) -> Result<Response<ResBody>, Erro
 			.body(string_body(xml))?)
 	} else {
 		Ok(Response::builder()
-			.status(StatusCode::NO_CONTENT)
+			.status(StatusCode::NOT_FOUND)
 			.body(empty_body())?)
 	}
 }
@@ -55,7 +53,6 @@ pub async fn handle_delete_lifecycle(ctx: ReqCtx) -> Result<Response<ResBody>, E
 pub async fn handle_put_lifecycle(
 	ctx: ReqCtx,
 	req: Request<ReqBody>,
-	content_sha256: Option<Hash>,
 ) -> Result<Response<ResBody>, Error> {
 	let ReqCtx {
 		garage,
@@ -64,11 +61,7 @@ pub async fn handle_put_lifecycle(
 		..
 	} = ctx;
 
-	let body = BodyExt::collect(req.into_body()).await?.to_bytes();
+	let body = req.into_body().collect().await?;
-
-	if let Some(content_sha256) = content_sha256 {
-		verify_signed_content(content_sha256, &body[..])?;
-	}
 
 	let conf: LifecycleConfiguration = from_reader(&body as &[u8])?;
 	let config = conf

src/api/s3/list.rs

@@ -13,13 +13,14 @@ use garage_model::s3::object_table::*;
 
 use garage_table::EnumerationOrder;
 
-use crate::encoding::*;
+use garage_api_common::encoding::*;
-use crate::helpers::*;
+use garage_api_common::helpers::*;
-use crate::s3::api_server::{ReqBody, ResBody};
+
-use crate::s3::encryption::EncryptionParams;
+use crate::api_server::{ReqBody, ResBody};
-use crate::s3::error::*;
+use crate::encryption::EncryptionParams;
-use crate::s3::multipart as s3_multipart;
+use crate::error::*;
-use crate::s3::xml as s3_xml;
+use crate::multipart as s3_multipart;
+use crate::xml as s3_xml;
 
 const DUMMY_NAME: &str = "Dummy Key";
 const DUMMY_KEY: &str = "GKDummyKey";
@@ -53,7 +54,6 @@ pub struct ListMultipartUploadsQuery {
 #[derive(Debug)]
 pub struct ListPartsQuery {
 	pub bucket_name: String,
-	pub bucket_id: Uuid,
 	pub key: String,
 	pub upload_id: String,
 	pub part_number_marker: Option<u64>,
@@ -398,7 +398,7 @@ enum ExtractionResult {
 		key: String,
 	},
 	// Fallback key is used for legacy APIs that only support
-	// exlusive pagination (and not inclusive one).
+	// exclusive pagination (and not inclusive one).
 	SkipTo {
 		key: String,
 		fallback_key: Option<String>,
@@ -408,7 +408,7 @@ enum ExtractionResult {
 #[derive(PartialEq, Clone, Debug)]
 enum RangeBegin {
 	// Fallback key is used for legacy APIs that only support
-	// exlusive pagination (and not inclusive one).
+	// exclusive pagination (and not inclusive one).
 	IncludingKey {
 		key: String,
 		fallback_key: Option<String>,
@@ -1244,10 +1244,8 @@ mod tests {
 
 	#[test]
 	fn test_fetch_part_info() -> Result<(), Error> {
-		let uuid = Uuid::from([0x08; 32]);
 		let mut query = ListPartsQuery {
 			bucket_name: "a".to_string(),
-			bucket_id: uuid,
 			key: "a".to_string(),
 			upload_id: "xx".to_string(),
 			part_number_marker: None,

src/api/s3/multipart.rs

@@ -1,13 +1,20 @@
 use std::collections::HashMap;
-use std::convert::TryInto;
+use std::convert::{TryFrom, TryInto};
+use std::hash::Hasher;
 use std::sync::Arc;
 
 use base64::prelude::*;
+use crc32c::Crc32cHasher as Crc32c;
+use crc32fast::Hasher as Crc32;
 use futures::prelude::*;
 use hyper::{Request, Response};
+use md5::{Digest, Md5};
+use sha1::Sha1;
+use sha2::Sha256;
 
 use garage_table::*;
 use garage_util::data::*;
+use garage_util::error::OkOrMessage;
 
 use garage_model::garage::Garage;
 use garage_model::s3::block_ref_table::*;
@@ -15,14 +22,14 @@ use garage_model::s3::mpu_table::*;
 use garage_model::s3::object_table::*;
 use garage_model::s3::version_table::*;
 
-use crate::helpers::*;
+use garage_api_common::helpers::*;
-use crate::s3::api_server::{ReqBody, ResBody};
+use garage_api_common::signature::checksum::*;
-use crate::s3::checksum::*;
+
-use crate::s3::encryption::EncryptionParams;
+use crate::api_server::{ReqBody, ResBody};
-use crate::s3::error::*;
+use crate::encryption::EncryptionParams;
-use crate::s3::put::*;
+use crate::error::*;
-use crate::s3::xml as s3_xml;
+use crate::put::*;
-use crate::signature::verify_signed_content;
+use crate::xml as s3_xml;
 
 // ----
 
@@ -42,7 +49,7 @@ pub async fn handle_create_multipart_upload(
 	let upload_id = gen_uuid();
 	let timestamp = next_timestamp(existing_object.as_ref());
 
-	let headers = get_headers(req.headers())?;
+	let headers = extract_metadata_headers(req.headers())?;
 	let meta = ObjectVersionMetaInner {
 		headers,
 		checksum: None,
@@ -93,7 +100,6 @@ pub async fn handle_put_part(
 	key: &str,
 	part_number: u64,
 	upload_id: &str,
-	content_sha256: Option<Hash>,
 ) -> Result<Response<ResBody>, Error> {
 	let ReqCtx { garage, .. } = &ctx;
 
@@ -104,17 +110,30 @@ pub async fn handle_put_part(
 			Some(x) => Some(x.to_str()?.to_string()),
 			None => None,
 		},
-		sha256: content_sha256,
+		sha256: None,
 		extra: request_checksum_value(req.headers())?,
 	};
 
-	// Read first chuck, and at the same time try to get object to see if it exists
 	let key = key.to_string();
 
-	let (req_head, req_body) = req.into_parts();
+	let (req_head, mut req_body) = req.into_parts();
-	let stream = body_stream(req_body);
+
+	// Before we stream the body, configure the needed checksums.
+	req_body.add_expected_checksums(expected_checksums.clone());
+	// TODO: avoid parsing encryption headers twice...
+	if !EncryptionParams::new_from_headers(&garage, &req_head.headers)?.is_encrypted() {
+		// For non-encrypted objects, we need to compute the md5sum in all cases
+		// (even if content-md5 is not set), because it is used as an etag of the
+		// part, which is in turn used in the etag computation of the whole object
+		req_body.add_md5();
+	}
+
+	let (stream, stream_checksums) = req_body.streaming_with_checksums();
+	let stream = stream.map_err(Error::from);
+
 	let mut chunker = StreamChunker::new(stream, garage.config.block_size);
 
+	// Read first chuck, and at the same time try to get object to see if it exists
 	let ((_, object_version, mut mpu), first_block) =
 		futures::try_join!(get_upload(&ctx, &key, &upload_id), chunker.next(),)?;
 
@@ -171,21 +190,21 @@ pub async fn handle_put_part(
 	garage.version_table.insert(&version).await?;
 
 	// Copy data to version
-	let checksummer =
+	let (total_size, _, _) = read_and_put_blocks(
-		Checksummer::init(&expected_checksums, !encryption.is_encrypted()).add(checksum_algorithm);
-	let (total_size, checksums, _) = read_and_put_blocks(
 		&ctx,
 		&version,
 		encryption,
 		part_number,
 		first_block,
-		&mut chunker,
+		chunker,
-		checksummer,
+		Checksummer::new(),
 	)
 	.await?;
 
-	// Verify that checksums map
+	// Verify that checksums match
-	checksums.verify(&expected_checksums)?;
+	let checksums = stream_checksums
+		.await
+		.ok_or_internal_error("checksum calculation")??;
 
 	// Store part etag in version
 	let etag = encryption.etag_from_md5(&checksums.md5);
@@ -247,7 +266,6 @@ pub async fn handle_complete_multipart_upload(
 	req: Request<ReqBody>,
 	key: &str,
 	upload_id: &str,
-	content_sha256: Option<Hash>,
 ) -> Result<Response<ResBody>, Error> {
 	let ReqCtx {
 		garage,
@@ -259,11 +277,7 @@ pub async fn handle_complete_multipart_upload(
 
 	let expected_checksum = request_checksum_value(&req_head.headers)?;
 
-	let body = http_body_util::BodyExt::collect(req_body).await?.to_bytes();
+	let body = req_body.collect().await?;
-
-	if let Some(content_sha256) = content_sha256 {
-		verify_signed_content(content_sha256, &body[..])?;
-	}
 
 	let body_xml = roxmltree::Document::parse(std::str::from_utf8(&body)?)?;
 	let body_list_of_parts = parse_complete_multipart_upload_body(&body_xml)
@@ -429,7 +443,16 @@ pub async fn handle_complete_multipart_upload(
 	// Send response saying ok we're done
 	let result = s3_xml::CompleteMultipartUploadResult {
 		xmlns: (),
-		location: None,
+		// FIXME: the location returned is not always correct:
+		// - we always return https, but maybe some people do http
+		// - if root_domain is not specified, a full URL is not returned
+		location: garage
+			.config
+			.s3_api
+			.root_domain
+			.as_ref()
+			.map(|rd| s3_xml::Value(format!("https://{}.{}/{}", bucket_name, rd, key)))
+			.or(Some(s3_xml::Value(format!("/{}/{}", bucket_name, key)))),
 		bucket: s3_xml::Value(bucket_name.to_string()),
 		key: s3_xml::Value(key),
 		etag: s3_xml::Value(format!("\"{}\"", etag)),
@@ -592,3 +615,99 @@ fn parse_complete_multipart_upload_body(
 
 	Some(parts)
 }
+
+// ====== checksummer ====
+
+#[derive(Default)]
+pub(crate) struct MultipartChecksummer {
+	pub md5: Md5,
+	pub extra: Option<MultipartExtraChecksummer>,
+}
+
+pub(crate) enum MultipartExtraChecksummer {
+	Crc32(Crc32),
+	Crc32c(Crc32c),
+	Sha1(Sha1),
+	Sha256(Sha256),
+}
+
+impl MultipartChecksummer {
+	pub(crate) fn init(algo: Option<ChecksumAlgorithm>) -> Self {
+		Self {
+			md5: Md5::new(),
+			extra: match algo {
+				None => None,
+				Some(ChecksumAlgorithm::Crc32) => {
+					Some(MultipartExtraChecksummer::Crc32(Crc32::new()))
+				}
+				Some(ChecksumAlgorithm::Crc32c) => {
+					Some(MultipartExtraChecksummer::Crc32c(Crc32c::default()))
+				}
+				Some(ChecksumAlgorithm::Sha1) => Some(MultipartExtraChecksummer::Sha1(Sha1::new())),
+				Some(ChecksumAlgorithm::Sha256) => {
+					Some(MultipartExtraChecksummer::Sha256(Sha256::new()))
+				}
+			},
+		}
+	}
+
+	pub(crate) fn update(
+		&mut self,
+		etag: &str,
+		checksum: Option<ChecksumValue>,
+	) -> Result<(), Error> {
+		self.md5
+			.update(&hex::decode(&etag).ok_or_message("invalid etag hex")?);
+		match (&mut self.extra, checksum) {
+			(None, _) => (),
+			(
+				Some(MultipartExtraChecksummer::Crc32(ref mut crc32)),
+				Some(ChecksumValue::Crc32(x)),
+			) => {
+				crc32.update(&x);
+			}
+			(
+				Some(MultipartExtraChecksummer::Crc32c(ref mut crc32c)),
+				Some(ChecksumValue::Crc32c(x)),
+			) => {
+				crc32c.write(&x);
+			}
+			(Some(MultipartExtraChecksummer::Sha1(ref mut sha1)), Some(ChecksumValue::Sha1(x))) => {
+				sha1.update(&x);
+			}
+			(
+				Some(MultipartExtraChecksummer::Sha256(ref mut sha256)),
+				Some(ChecksumValue::Sha256(x)),
+			) => {
+				sha256.update(&x);
+			}
+			(Some(_), b) => {
+				return Err(Error::internal_error(format!(
+					"part checksum was not computed correctly, got: {:?}",
+					b
+				)))
+			}
+		}
+		Ok(())
+	}
+
+	pub(crate) fn finalize(self) -> (Md5Checksum, Option<ChecksumValue>) {
+		let md5 = self.md5.finalize()[..].try_into().unwrap();
+		let extra = match self.extra {
+			None => None,
+			Some(MultipartExtraChecksummer::Crc32(crc32)) => {
+				Some(ChecksumValue::Crc32(u32::to_be_bytes(crc32.finalize())))
+			}
+			Some(MultipartExtraChecksummer::Crc32c(crc32c)) => Some(ChecksumValue::Crc32c(
+				u32::to_be_bytes(u32::try_from(crc32c.finish()).unwrap()),
+			)),
+			Some(MultipartExtraChecksummer::Sha1(sha1)) => {
+				Some(ChecksumValue::Sha1(sha1.finalize()[..].try_into().unwrap()))
+			}
+			Some(MultipartExtraChecksummer::Sha256(sha256)) => Some(ChecksumValue::Sha256(
+				sha256.finalize()[..].try_into().unwrap(),
+			)),
+		};
+		(md5, extra)
+	}
+}